利用隐式的用户反馈来提高互动视频检索</t我tle> <link href="//www.newsama.com/static/themes/hindawi/base.css" rel="stylesheet"> <link href="https://static-01.hindawi.com/next_assets/wyT4heYybLmt1UEJexPE9/static/lib/basictable.css" rel="stylesheet" media="none" class="next-head"> <link rel="stylesheet" href="https://static-01.hindawi.com/next_assets/wyT4heYybLmt1UEJexPE9/static/lib/owl.carousel.min.css" media="none" class="next-head"> <link rel="stylesheet" href="https://static-01.hindawi.com/next_assets/wyT4heYybLmt1UEJexPE9/static/lib/owl.theme.default.min.css" media="none" class="next-head"> <link rel="preload" type="text/css" href="https://cdn.bibblio.org/rcm/4.24/bib-related-content.min.css" as="style" class="next-head"> <link rel="preload" type="text/css" href="https://cdn.bibblio.org/rcm/4.24/bib-related-content.min.css" as="style" class="next-head"> <link rel="stylesheet" href="https://static-01.hindawi.com/next_assets/wyT4heYybLmt1UEJexPE9/_next/static/css/common.835aedbf.chunk.css"> <link rel="stylesheet" href="https://static-01.hindawi.com/next_assets/wyT4heYybLmt1UEJexPE9/_next/static/css/static/wyT4heYybLmt1UEJexPE9/pages/journal_article.js.6cbeea55.chunk.css"> <style data-styled="NQpXC gnduKR jzXSTA bNmdOy epRoim bUhGXt gjctdt jGJLow kAzjOl ivbKJD fopjOz hbVEoI khbeDd gvAGTV kNoiZu diutvq bqeMOy bMdQcY ldSkVQ cmzomn fJLQxX kuxTps prcY bJcSpM elhyhe fTkdCR jCfZqq EwsTm hYnXNL hGCBfz dfaOeH jOLhQl iiusBg TsaZF cqhPZs gaRePw dEfJbW eBepOs fSPaCp bVEEaH etauus ctePXh gCIfTO hHKuAX dIjoCg jUTgTa eujois cQgXqt ewCAZX kdvWPs hAFFGk lkWWFY ehYbwJ legory epgILe kCfRMJ bjwAQR byflLi jXeALb UVThf dDyxFF fkVFlF dzZrgS jLUVhb dVCLgr bBmZJZ ewHMLz bbzREB jknsZc eSvjPB jntdRu hAzjHP mBGSe bkiEfx ztvYg cUsFHx duuwge lcOiDF gsmOCv kPshFw kEVvRd jZOxOO eHMFfo XkrFe irNWJI iQSObO dMuGGA kfDwZY evMkvU fYrLtk fjviok hgKieC cIribm khHUcu gvEcCq dbAANp eyHfWh cBTaXT becpDt hFGNjP jzTMLK gbrQxi hRnXob hymroZ cbcyj hmDWcT isgRZA ihmyEq itFKcQ cGuwDn jrsDZP gWTCXQ jGPvPL etzuBr gWgCNb WHkfb ibWUzs kdRRST cHatnS kxdGAZ jHPgUP fXBKFa hNPrIr bnbfTS cYsVdk inBNwS dkutRV hlGpbO filPdg ecHxGW gCiATD eEFYIL dsXEHZ iXKuYd" data-styled-version="4.4.1"> /* sc-component-id: sc-bdVaJa */ .EwsTm{height:200vh;width:100%;position:fixed;z-index:1;background:rgba(255,255,255,0.8);} .EwsTm.isHide{display:none;} @media (min-width:768px) and (max-width:991px){} @media (min-width:992px){.EwsTm{display:none;}} /* sc-component-id: sc-htpNat */ .bUhGXt{color:black;-webkit-text-decoration:none;text-decoration:none;border-radius:5px;font-size:16px;} .bUhGXt:focus-visible{outline-width:3px;outline-color:#8fc9ff;outline-style:solid;} /* sc-component-id: sc-bxivhb */ .ivbKJD{display:inline-block;} /* sc-component-id: sc-ifAKCX */ .fopjOz{display:-webkit-inline-box;display:-webkit-inline-flex;display:-ms-inline-flexbox;display:inline-flex;-webkit-flex:wrap;-ms-flex:wrap;flex:wrap;-webkit-align-items:center;-webkit-box-align:center;-ms-flex-align:center;align-items:center;list-style-type:none;margin:0;padding:0;} .fopjOz li{padding:0 10px;} .fopjOz li a{padding:0;} .fopjOz a{-webkit-transition:border .25s ease-in-out 0s;transition:border .25s ease-in-out 0s;border-bottom:2px solid #fff;border-radius:0;} .fopjOz a:hover{border-bottom:2px solid #9b9b9b;padding-bottom:4px;} .fopjOz .active{border-bottom:2px solid #57a61e;} /* sc-component-id: sc-EHOje */ .jOLhQl{vertical-align:middle;height:24px;}.iiusBg{vertical-align:middle;height:16px;}.TsaZF{vertical-align:middle;height:40px;} /* sc-component-id: sc-bZQynM */ .bNmdOy{max-width:1440px;margin:0 auto;width:100%;} /* sc-component-id: sc-htoDjs */ .khbeDd{vertical-align:middle;font-size:24px;}.gvAGTV{vertical-align:middle;font-size:9px;}.kNoiZu{vertical-align:middle;font-size:16px;}.diutvq{vertical-align:middle;font-size:22px;}.bqeMOy{vertical-align:middle;font-size:20px;}.bMdQcY{vertical-align:middle;font-size:11px;} /* sc-component-id: sc-fjdhpX */ .jzXSTA{position:fixed;top:0;left:0;right:0;z-index:1000;background:#fff;height:70px;font-size:16px;font-weight:500;border-bottom:1px solid #e3e3e3;-webkit-transition:all 0.5s;transition:all 0.5s;} @media (max-width:991px){.jzXSTA{border-bottom:1px solid #e3e3e3;}} @media (max-width:767px){.jzXSTA{margin-bottom:70px;height:60px;}} .jzXSTA .header__padding{padding:0 20px;box-sizing:border-box;} @media (max-width:991px){.jzXSTA .header__padding{padding:0 0 0 20px;}} .jzXSTA .menu{position:fixed;top:60px;right:0;width:100%;background:white;} @media (max-width:991px){.jzXSTA .menu{z-index:2;}} @media (min-width:768px){.jzXSTA .menu{width:500px;top:70px;}} @media (min-width:991.5px){.jzXSTA .menu{display:none;}} .jzXSTA .menu ul{list-style-type:none;text-align:center;padding:0;margin:0;} @media (min-width:768px){.jzXSTA .menu ul{padding:0;border-left:1px solid #e3e3e3;}} .jzXSTA .menu ul li a{color:#4a4a4a;padding:10px;-webkit-text-decoration:none;text-decoration:none;display:block;border-bottom:1px solid #e3e3e3;} .jzXSTA .menu .button_wrap{width:90%;margin:10px auto;border-radius:5px;border-style:solid;border-width:1px;} @media (min-width:768px){.jzXSTA .menu .button_wrap{display:none;}} .jzXSTA .menu .button__publishing{background-color:#6ba439;color:#fff;border-color:#6ba439;} .jzXSTA .menu .button__login{display:none;background-color:#fff;color:#4D8A17;border-color:#e3e3e3;} .jzXSTA.headerUp{-webkit-transform:translateY(-100%);-ms-transform:translateY(-100%);transform:translateY(-100%);} /* sc-component-id: sc-jzJRlG */ .jGJLow{-webkit-box-pack:end;-webkit-justify-content:flex-end;-ms-flex-pack:end;justify-content:flex-end;} .jGJLow a{padding:10px;} .jGJLow nav{-webkit-align-self:center;-ms-flex-item-align:center;align-self:center;} @media (max-width:991px){.jGJLow nav{display:none;}} .jGJLow nav ul{margin-right:16px;} .jGJLow nav ul li a{color:#4a4a4a;padding:10px 0px 4px;margin:0 6px;} @media (max-width:991px){.jGJLow .icon-themeable{border-left:1px solid #e3e3e3;padding:23px 20px;}} @media (max-width:767px){.jGJLow .icon-themeable{padding:18px 20px;}} @media (max-width:767.5px){.jGJLow.mobileButtonFlex{-ms-flex:0 0 51%;-webkit-flex:0 0 51%;-ms-flex:0 0 51%;flex:0 0 51%;max-width:51%;}} /* sc-component-id: sc-cSHVUG */ .ldSkVQ{display:none;padding:0;background-color:#fff;border:none;height:59px;} @media (min-width:768px){.ldSkVQ{height:69px;}} .ldSkVQ img{box-sizing:content-box;width:28px;padding:17px 20px;} @media (min-width:768px){.ldSkVQ img{padding:22px 20px;}} @media (max-width:991px){.ldSkVQ{display:inline-block;}} /* sc-component-id: sc-kAzzGY */ .epRoim a{display:inline-block;padding:15px 0;} @media (max-width:767px){.epRoim a{padding:10px 0;}} @media (max-width:767.5px){.epRoim.mobileLogoFlex{-ms-flex:0 0 49%;-webkit-flex:0 0 49%;-ms-flex:0 0 49%;flex:0 0 49%;max-width:49%;}} /* sc-component-id: sc-chPdSV */ .hbVEoI{border:none;height:50px;background-color:#fff;margin:10px 0 0 17px;position:relative;} @media (max-width:991.5px){.hbVEoI{margin:0 3px 0 16px;padding:0;height:69px;}} @media (max-width:767.5px){.hbVEoI{height:59px;}} .hbVEoI img{box-sizing:content-box;} .hbVEoI .icon__cross{width:24px;height:24px;} @media (min-width:768px){.hbVEoI .icon__cross{padding:23px 20px;}} @media (min-width:992px){.hbVEoI .icon__cross{padding:0px;}} /* sc-component-id: sc-ckVGcZ */ .gjctdt{vertical-align:middle;height:39px;} @media (max-width:401px){.gjctdt{max-width:100%;}} /* sc-component-id: sc-jKJlTe */ .inBNwS{background-color:#0d2934;font-size:16px;position:relative;z-index:3;} @media (max-width:991px){.inBNwS{z-index:0;}} /* sc-component-id: sc-eNQAEJ */ .dkutRV{border-bottom :1px solid;border-color:#828282;padding:29px 40px 30px 40px;} @media (min-width:992px){.dkutRV{padding:40px 40px 40px 35px;}} .dkutRV .primary__footer .mobile{display:block;} .dkutRV .primary__footer .desktop{display:none;} @media (min-width:992px){.dkutRV .primary__footer{-webkit-box-pack:justify;-webkit-justify-content:space-between;-ms-flex-pack:justify;justify-content:space-between;}.dkutRV .primary__footer .mobile{display:none;}.dkutRV .primary__footer .desktop{display:block;}} @media (min-width:992px){.dkutRV .col5{-webkit-flex:0 0 18%;-ms-flex:0 0 18%;flex:0 0 18%;max-width:18%;}} .dkutRV address{font-style:inherit;line-height:1.5;font-size:16px;display:none;border-right:1px solid #EBEBEB;} @media (max-width:767px){.dkutRV address{border-bottom:1px solid #E3E3E3;padding-bottom:20px;}} @media (min-width:992px){.dkutRV address{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-flex-direction:column;-ms-flex-direction:column;flex-direction:column;-webkit-box-pack:center;-webkit-justify-content:center;-ms-flex-pack:center;justify-content:center;margin-right:50px;}} @media (min-width:1200px){.dkutRV address{margin-right:70px;}} .dkutRV address p{margin:0;font-size:14px;} @media (max-width:767px){.dkutRV address p{text-align:center;}} @media (min-width:1200px){.dkutRV address p{font-size:16px;}} .dkutRV nav ul{list-style:none;padding:0;line-height:2;font-size:16px;margin:18px 0 0 0;} .dkutRV nav ul:first-child{margin:0;} @media (min-width:992px){.dkutRV nav ul{margin:0;}} .dkutRV nav ul li{line-height:22px;padding:5px;} .dkutRV nav ul li a{color:#d8d8d8;font-weight:500;display:block;font-size:16px;} .dkutRV nav ul li img{padding:2.5px 0 0 10px;box-sizing:content-box;-webkit-transform:rotate(180deg);-ms-transform:rotate(180deg);transform:rotate(180deg);} /* sc-component-id: sc-hMqMXs */ .hlGpbO{padding-top:6px;color:#FFFFFF;font-weight:500;margin-bottom:29px;} @media (min-width:992px){.hlGpbO{-webkit-flex:0 0 25%;-ms-flex:0 0 25%;flex:0 0 25%;max-width:-webkit-fit-content;max-width:-moz-fit-content;max-width:fit-content;}} /* sc-component-id: sc-kEYyzF */ .ecHxGW{margin-right:20px;} .ecHxGW:last-child{margin-right:0;} /* sc-component-id: sc-kkGfuU */ .filPdg{margin-top:15px;} /* sc-component-id: sc-iAyFgw */ .gCiATD{line-height:30px;font-size:11px;padding:10px 20px;text-align:center;} .gCiATD .textright{text-align:right;font-weight:500;} @media (max-width:767px){.gCiATD .textright{text-align:center;}} @media (min-width:992px){.gCiATD .textright{font-weight:100;}} .gCiATD a{font-size:13px;color:#fff;} /* sc-component-id: sc-hSdWYo */ .etauus{display:inline-block;background-color:#fff;color:#4D8A17;border-color:#E3E3E3;width:auto;border-radius:5px;font-size:16px;border-width:1px;border-style:solid;} .etauus:focus-visible{outline-width:3px;outline-color:#8fc9ff;outline-style:solid;} /* sc-component-id: sc-eHgmQL */ .fTkdCR{font-size:22px;color:#000;} @media (min-width:767.5px){.fTkdCR{font-size:24px;}} @media (min-width:991.5px){.fTkdCR{font-size:25px;}} .fTkdCR:after{content:'';display:inline-block;margin-top:0.2em;margin-left:0.5em;} /* sc-component-id: sc-iRbamj */ .jZOxOO{font-size:30px;font-weight:500;-webkit-letter-spacing:1px;-moz-letter-spacing:1px;-ms-letter-spacing:1px;letter-spacing:1px;line-height:1;margin-top:0;} @media (min-width:767.5px){.jZOxOO{font-size:40px;}} @media (min-width:991.5px){.jZOxOO{font-size:40px;}} .jZOxOO sup,.jZOxOO sub{vertical-align:baseline;position:relative;top:-10px;} .jZOxOO sub{top:10px;} /* sc-component-id: sc-Rmtcm */ .hNPrIr{font-size:20px;font-weight:500;margin-top:0;margin-bottom:15px;} /* sc-component-id: sc-bRBYWo */ .fXBKFa{display:inline-block;} @media (max-width:991.5px){.fXBKFa{width:100%;}} @media (min-width:991.5px){.fXBKFa.hide{display:none !important;}} /* sc-component-id: sc-hzDkRC */ .bnbfTS{display:-webkit-inline-box;display:-webkit-inline-flex;display:-ms-inline-flexbox;display:inline-flex;-webkit-flex:wrap;-ms-flex:wrap;flex:wrap;-webkit-align-items:center;-webkit-box-align:center;-ms-flex-align:center;align-items:center;list-style-type:none;margin:0;width:100%;padding:0 !important;border-top:1px solid #fff;} .bnbfTS li{width:100%;text-align:center;box-sizing:border-box;padding:17px 0px;min-height:48px;} .bnbfTS li a{display:block;} .bnbfTS .iconBorder{border-right:1px solid #fff;} .bnbfTS img{width:23px;} .bnbfTS img.socialLinkedin{width:28px;} .bnbfTS .wechat-container{position:relative;} .bnbfTS .wechat-container .addthis_inline_share_toolbox{position:absolute;top:0;left:0;width:100%;height:100%;opacity:0 !important;} /* sc-component-id: sc-jtRfpW */ .hHKuAX{background-color:#6BA439;color:#fff;padding:20px;line-height:22px !important;border-radius:0;border-top:1px solid #fff;display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-box-pack:justify;-webkit-justify-content:space-between;-ms-flex-pack:justify;justify-content:space-between;min-height:60px;} .hHKuAX img{display:-webkit-inline-box;display:-webkit-inline-flex;display:-ms-inline-flexbox;display:inline-flex;-webkit-align-self:center;-ms-flex-item-align:center;align-self:center;} .hHKuAX span{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-align-self:center;-ms-flex-item-align:center;align-self:center;} .hHKuAX span.text{padding-left:20px;font-weight:500;} @media (max-width:360px){.hHKuAX span.text{padding-left:5px;font-size:14px;}} @media (min-width:320px) and (max-width:767.5px){.hHKuAX span.text{padding-left:20px;}}.dIjoCg{background-color:#F2F2F2;color:#4A4A4A;padding:20px;line-height:22px !important;border-radius:0;border-top:1px solid #E3E3E3;display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-box-pack:justify;-webkit-justify-content:space-between;-ms-flex-pack:justify;justify-content:space-between;min-height:60px;} .dIjoCg img{display:-webkit-inline-box;display:-webkit-inline-flex;display:-ms-inline-flexbox;display:inline-flex;-webkit-align-self:center;-ms-flex-item-align:center;align-self:center;} .dIjoCg span{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-align-self:center;-ms-flex-item-align:center;align-self:center;} .dIjoCg span.text{padding-left:20px;font-weight:500;} @media (max-width:360px){.dIjoCg span.text{padding-left:5px;font-size:14px;}} @media (min-width:320px) and (max-width:767.5px){.dIjoCg span.text{padding-left:20px;}} /* sc-component-id: sc-kTUwUJ */ .eujois{width:20px;} /* sc-component-id: sc-dqBHgY */ .cQgXqt{width:20px;height:16px;color:#6ba439;} /* sc-component-id: sc-cbkKFq */ .kuxTps{-webkit-transition:-webkit-transform 0.5s;-webkit-transition:transform 0.5s;transition:transform 0.5s;position:fixed;top:60px;} @media (min-width:767.5px){.kuxTps{position:fixed;top:70px;}} @media (min-width:991.5px){.kuxTps{position:fixed;top:70px;}}.prcY{-webkit-transition:-webkit-transform 0.5s;-webkit-transition:transform 0.5s;transition:transform 0.5s;top:0px;} @media (min-width:767.5px){.prcY{top:0px;}} @media (min-width:991.5px){.prcY{position:fixed;top:241px;}} /* sc-component-id: sc-hGoxap */ .dEfJbW{position:relative;background:#007d8d;} @media (max-width:991px){.dEfJbW{z-index:2;}} @media (min-width:767.5px) and (max-width:991.5px){.dEfJbW{width:50%;}} @media (min-width:991.5px){.dEfJbW{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-box-pack:end;-webkit-justify-content:flex-end;-ms-flex-pack:end;justify-content:flex-end;border-top:1px solid white;}} .dEfJbW.isHide{display:none;} @media (min-width:991.5px){.dEfJbW.isHide{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;}} .dEfJbW .submit__button{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;box-shadow:0 2px 4px rgba(0,0,0,.5);padding:0 20px;min-height:45px;} .dEfJbW .submit__button span.text{font-size:16px;} .dEfJbW .submit__button img{width:20px;} @media (min-width:767.5px){.dEfJbW .submit__button{min-height:58px;}.dEfJbW .submit__button img{width:22px;}} @media (min-width:991.5px){.dEfJbW .submit__button{display:none;}} /* sc-component-id: sc-fjmCvl */ .fSPaCp{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;width:100%;font-size:16px;font-weight:500;color:#fff;background-color:#007d8d;line-height:20px;padding:10px 20px;border:0;border-radius:0;border-top:1px solid #fff;-webkit-box-pack:justify;-webkit-justify-content:space-between;-ms-flex-pack:justify;justify-content:space-between;margin:0;} @media (min-width:767.5px){.fSPaCp{padding:16px 20px;}} @media (min-width:991.5px){.fSPaCp{border-top:none;border-right:1px solid #fff;padding:14.5px 15px;font-size:14px;}} @media (min-width:1200px){.fSPaCp{padding:14.5px 20px;font-size:16px;}} .fSPaCp .icon__inactive{-webkit-transform:rotateZ(-90deg);-ms-transform:rotateZ(-90deg);transform:rotateZ(-90deg);line-height:20px;} @media (min-width:991.5px){.fSPaCp .icon__inactive{-webkit-transform:rotateZ(0);-ms-transform:rotateZ(0);transform:rotateZ(0);}} .fSPaCp.button__active{background:#fff;color:#4A4A4A;} @media (min-width:767.5px) and (max-width:991.5px){.fSPaCp.button__active{border-right:1px solid #007f8f;}} .fSPaCp .icon__active{-webkit-transform:rotateZ(-90deg);-ms-transform:rotateZ(-90deg);transform:rotateZ(-90deg);line-height:20px;} @media (min-width:991.5px){.fSPaCp .icon__active{-webkit-transform:rotate(0);-ms-transform:rotate(0);transform:rotate(0);}} .fSPaCp .icon__hidden{display:none;}.bVEEaH{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;width:100%;font-size:16px;font-weight:500;color:#fff;background-color:#007d8d;line-height:20px;padding:10px 20px;border:0;border-radius:0;border-top:1px solid #fff;-webkit-box-pack:justify;-webkit-justify-content:space-between;-ms-flex-pack:justify;justify-content:space-between;margin:0;} @media (min-width:767.5px){.bVEEaH{padding:16px 20px;}} @media (min-width:991.5px){.bVEEaH{border-top:none;border-right:0;padding:14.5px 15px;font-size:14px;}} @media (min-width:1200px){.bVEEaH{padding:14.5px 20px;font-size:16px;}} .bVEEaH .icon__inactive{-webkit-transform:rotateZ(-90deg);-ms-transform:rotateZ(-90deg);transform:rotateZ(-90deg);line-height:20px;} @media (min-width:991.5px){.bVEEaH .icon__inactive{-webkit-transform:rotateZ(0);-ms-transform:rotateZ(0);transform:rotateZ(0);}} .bVEEaH.button__active{background:#fff;color:#4A4A4A;} @media (min-width:767.5px) and (max-width:991.5px){.bVEEaH.button__active{border-right:1px solid #007f8f;}} .bVEEaH .icon__active{-webkit-transform:rotateZ(-90deg);-ms-transform:rotateZ(-90deg);transform:rotateZ(-90deg);line-height:20px;} @media (min-width:991.5px){.bVEEaH .icon__active{-webkit-transform:rotate(0);-ms-transform:rotate(0);transform:rotate(0);}} .bVEEaH .icon__hidden{display:none;} /* sc-component-id: sc-TFwJa */ .gCIfTO{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;width:100%;font-size:16px;font-weight:500;color:#fff;background-color:#007d8d;line-height:20px;padding:10px 20px;border:0;border-radius:0;border-top:1px solid #fff;-webkit-box-pack:justify;-webkit-justify-content:space-between;-ms-flex-pack:justify;justify-content:space-between;} @media (min-width:767.5px){.gCIfTO{padding:16px 20px;}} @media (min-width:991.5px){.gCIfTO{-webkit-transition:all 0.25s;transition:all 0.25s;border-top:none;border-right:1px solid #fff;border-left:0;padding:14.5px 30px 14.5px 20px;width:calc(100% / 6);font-size:14px;}.gCIfTO:hover{background:#fff;color:#4A4A4A;}} @media (min-width:1200px){.gCIfTO{font-size:16px;}} /* sc-component-id: sc-bHwgHz */ .ctePXh{width:18px;margin-right:-4px;} @media (min-width:991.5px){.ctePXh{width:7px;font-size:7px;}} @media (min-width:1200px){.ctePXh{width:18px;}} /* sc-component-id: sc-krDsej */ .ewCAZX{position:absolute;top:0;width:calc(100% - 20px);right:0;bottom:0;background:#fff;-webkit-transition:-webkit-transform 0.5s;-webkit-transition:transform 0.5s;transition:transform 0.5s;min-height:290px;box-shadow:0 2px 4px rgba(0,0,0,0.5);} .ewCAZX.hide__submenu{-webkit-transform:translateX(100%);-ms-transform:translateX(100%);transform:translateX(100%);} @media (min-width:767.5px){.ewCAZX.hide__submenu{-webkit-transform:translateX(-200%);-ms-transform:translateX(-200%);transform:translateX(-200%);}} @media (min-width:767.5px){.ewCAZX{left:100%;width:100%;padding-top:20px;z-index:-1;min-height:375px;}} @media (min-width:991.5px){.ewCAZX{display:none;}} /* sc-component-id: sc-dTdPqK */ .lkWWFY{-webkit-transition:-webkit-transform 0.5s;-webkit-transition:transform 0.5s;transition:transform 0.5s;display:none;} @media (min-width:991.5px){.lkWWFY.hide__submenu{-webkit-transform:translateY(-300%);-ms-transform:translateY(-300%);transform:translateY(-300%);}} @media (min-width:991.5px){.lkWWFY{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;height:186px;top:49px;left:0;position:absolute;right:0;background:#fff;z-index:-1;box-shadow:0 4px 6px rgba(0,0,0,0.21);}} /* sc-component-id: sc-itybZL */ .kdvWPs{font-weight:700;border:0;color:#4A4A4A;padding:10px 0;background-color:#fff;font-size:16px;line-height:20px;} .kdvWPs:focus-visible{outline-width:3px;outline-color:#8fc9ff;outline-style:solid;} @media (min-width:767.5px){.kdvWPs{display:none;}} /* sc-component-id: sc-fzsDOv */ .hAFFGk{margin:0px 10px 2px 20px;-webkit-transform:rotate(90deg);-ms-transform:rotate(90deg);transform:rotate(90deg);display:inline-block;} /* sc-component-id: sc-gPWkxV */ .legory{width:50%;} /* sc-component-id: sc-jVODtj */ .epgILe{color:#007B8B;width:50%;font-size:16px;line-height:26px;margin:auto;padding:0 6%;} /* sc-component-id: sc-kUaPvJ */ .ehYbwJ{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;width:50%;padding:24.5px 2% 0;-webkit-order:0;-ms-flex-order:0;order:0;} @media (min-width:1024px){.ehYbwJ{padding:24.5px 3% 0;}} /* sc-component-id: sc-giadOv */ .eBepOs{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;width:100%;} @media (min-width:991.5px){.eBepOs{width:calc(100% / 6);}.eBepOs:first-child{border-left:none;}} /* sc-component-id: sc-fONwsr */ .kCfRMJ{background-color:#D8D8D8;position:absolute;left:50%;display:inline-block;height:121px;width:2px;top:33px;} /* sc-component-id: sc-LKuAh */ .lcOiDF{padding:0;margin:10px 0 25px 0;} /* sc-component-id: sc-iBEsjs */ .kPshFw{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;border:none;font-size:12px;font-weight:500;margin-top:7px;padding:0;color:#4D8A17;-webkit-text-decoration:none;text-decoration:none;cursor:pointer;} .kPshFw:hover{-webkit-text-decoration:underline;text-decoration:underline;} @media (min-width:767.5px){.kPshFw{margin-top:10px;}} /* sc-component-id: sc-hzNEM */ .gsmOCv{overflow-y:hidden;height:auto;max-height:100%;-webkit-transition:all .5s;transition:all .5s;background-color:#F2F2F2;padding:0 16px;} .gsmOCv.isHide{max-height:0;-webkit-transition:all .5s;transition:all .5s;} .gsmOCv p{font-size:14px;line-height:22px;} .gsmOCv sup{margin-top:-5px;} /* sc-component-id: sc-chbbiW */ .mBGSe{padding:0;} /* sc-component-id: sc-kZmsYB */ .bkiEfx{font-size:14px;font-weight:700;line-height:16px;-webkit-letter-spacing:0.58px;-moz-letter-spacing:0.58px;-ms-letter-spacing:0.58px;letter-spacing:0.58px;margin:23px 0 0px !important;} @media (min-width:767.5px){.bkiEfx{margin:34px 0 0px !important;}} @media (min-width:991.5px){.bkiEfx{margin:-4px 0 0 !important;}} /* sc-component-id: sc-RcBXQ */ .ztvYg{font-size:14px;line-height:22px;} .ztvYg a{color:#4D8A17;-webkit-text-decoration:none;text-decoration:none;cursor:pointer;} .ztvYg a:hover{-webkit-text-decoration:underline;text-decoration:underline;} /* sc-component-id: sc-iSDuPN */ @font-face{font-family:'STIXGeneral-Bold';src:url('https://static-01.hindawi.com/next_assets/wyT4heYybLmt1UEJexPE9/_next/static/node_modules/hindawi-ui/fonts/STIXGeneral-Bold.otf');font-weight:normal;font-style:normal;} .kEVvRd{font-weight:700;font-family:STIXGeneral-Bold;margin:19px 0 22px;font-size:25px;line-height:28px;-webkit-letter-spacing:normal;-moz-letter-spacing:normal;-ms-letter-spacing:normal;letter-spacing:normal;} @media (min-width:767.5px){.kEVvRd{font-size:25px;margin:20px 0 23px;}} @media (min-width:991.5px){.kEVvRd{font-size:30px;line-height:34px;margin:19px 0 22px;}} .kEVvRd p{margin:0;} /* sc-component-id: sc-fZwumE */ .cUsFHx{padding:0 3px;} /* sc-component-id: sc-fQejPQ */ .duuwge{font-size:14px;} .duuwge:hover{-webkit-text-decoration:underline;text-decoration:underline;} /* sc-component-id: sc-jtRlXQ */ @font-face{font-family:'STIXGeneral-Regular';src:url('https://static-01.hindawi.com/next_assets/wyT4heYybLmt1UEJexPE9/_next/static/node_modules/hindawi-ui/fonts/STIXGeneral-Regular.otf');font-weight:normal;font-style:normal;} @font-face{src:url('https://static-01.hindawi.com/next_assets/wyT4heYybLmt1UEJexPE9/_next/static/node_modules/hindawi-ui/fonts/STIXGeneral-Regular.otf');font-weight:normal;font-style:normal;} @font-face{font-family:'STIXGeneral-Regular';src:url('https://static-01.hindawi.com/next_assets/wyT4heYybLmt1UEJexPE9/_next/static/node_modules/hindawi-ui/fonts/STIXGeneral-Regular.otf');font-weight:normal;font-style:normal;} .gvEcCq{font-family:STIXGeneral-Regular;text-align:justify;} .gvEcCq a{word-break:break-all;} .gvEcCq h4:first-of-type{border-top:0px solid #E3E3E3;padding-top:0;} .gvEcCq h4{font-size:20px;font-family:"IBM Plex Sans",sans-serif;border-top:1px solid #E3E3E3;padding-top:23px;margin-top:15px;margin-bottom:-3px;} .gvEcCq h4#abstract{margin-top:-7px;margin-bottom:-3px;} @media (min-width:767.5){.gvEcCq h4#abstract{margin-top:-6px;}} @media (min-width:991.5px){.gvEcCq h4#abstract{margin-top:34px;}} .gvEcCq h4.references,.gvEcCq h4.copyright{border-top:1px solid #E3E3E3;padding-top:23px;} .gvEcCq h4.references{margin-bottom:-8px;} .gvEcCq h5,.gvEcCq h6{margin:22px 0;font-size:16px;word-break:break-word;} .gvEcCq h6{margin:25px 0;text-align:left;} .gvEcCq p{font-size:16px;word-break:break-word;} .gvEcCq .floats-partial-section{border:1px solid #E3E3E3;margin:25px auto;} @media (max-width:767px){.gvEcCq .floats-partial-section{margin:0 auto;overflow-x:hidden;}} .gvEcCq .floats-partial-section .floats-partial-content{padding:19px 0px 19px 0px;height:254px;width:96%;overflow:hidden;margin:0 auto;position:relative;cursor:pointer;} .gvEcCq .floats-partial-section .floats-partial-content table{margin:0 auto;} .gvEcCq .floats-partial-section .floats-partial-content .table-group{width:-webkit-fit-content;width:-moz-fit-content;width:fit-content;} .gvEcCq .floats-partial-section .floats-partial-content .table-group table{width:-webkit-fit-content;width:-moz-fit-content;width:fit-content;} .gvEcCq .floats-partial-section .floats-partial-content .table-group table object{z-index:-1;position:relative;} .gvEcCq .floats-partial-section .floats-partial-content .table-group table td{padding:4px 7px;} .gvEcCq .floats-partial-section .floats-partial-content .table-group table td.align_left{text-align:left;} .gvEcCq .floats-partial-section .floats-partial-content .table-group table td.align_center{text-align:center;} @media (min-width:768px){.gvEcCq .floats-partial-section .floats-partial-content{padding:22px 0px 0 0px;height:197px;}} @media (min-width:992px){.gvEcCq .floats-partial-section .floats-partial-content{padding:23px 0px 13px 0px;height:215px;}} .gvEcCq .floats-partial-section .floats-partial-content:after{content:"";height:20%;position:absolute;bottom:0;right:0;left:0;background:-moz-linear-gradient(left,rgba(255,255,255,.7) 0%,rgba(255,255,255,.7) 56%,rgba(255,255,255,.7) 100%);background:-webkit-gradient(linear,left top,right top,color-stop(0%,rgba(255,255,255,.7)),color-stop(56%,rgba(255,255,255,.7)),color-stop(100%,rgba(255,255,255,.7)));background:-webkit-linear-gradient(rgba(255,255,255,0) 0%,rgba(255,255,255,0.5) 30%,rgba(255,255,255,0.7) 56%,rgba(255,255,255,0.9) 75%,rgba(255,255,255,1) 100%);background:-o-linear-gradient(left,rgba(255,255,255,.7) 0%,rgba(255,255,255,.7) 56%,rgba(255,255,255,.7) 100%);background:-ms-linear-gradient(left,rgba(255,255,255,.7) 0%,rgba(255,255,255,.7) 56%,rgba(255,255,255,.7) 100%);background:linear-gradient(left,rgba(255,255,255,.7) 0%,rgba(255,255,255,.7) 56%,rgba(255,255,255,.7) 100%);-webkit-filter:progid:DXImageTransform.Microsoft.gradient( startColorstr='#00ffffff',endColorstr='#ffffff',GradientType=1 );filter:progid:DXImageTransform.Microsoft.gradient( startColorstr='#00ffffff',endColorstr='#ffffff',GradientType=1 );} .gvEcCq .floats-partial-section .floats-partial-footer{background-color:#F2F2F2;border-top:1px solid #E3E3E3;} .gvEcCq .floats-partial-section .floats-partial-footer .floats-partial-caption{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-box-pack:end;-webkit-justify-content:flex-end;-ms-flex-pack:end;justify-content:flex-end;} .gvEcCq .floats-partial-section .floats-partial-footer .floats-partial-caption .minimize_btn{-webkit-flex:1;-ms-flex:1;flex:1;background:url('data:image/svg+xml;base64,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') no-repeat 6px;background-position:center;margin-top:10px;min-width:55px;} @media (min-width:768px){.gvEcCq .floats-partial-section .floats-partial-footer .floats-partial-caption .minimize_btn{display:none;}} .gvEcCq .floats-partial-section .floats-partial-footer .floats-partial-caption .expand_btn{-webkit-flex:1;-ms-flex:1;flex:1;background:url('data:image/svg+xml;base64,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') no-repeat 6px;background-position:center;margin-top:10px;min-width:55px;display:none;} @media (min-width:768px){.gvEcCq .floats-partial-section .floats-partial-footer .floats-partial-caption .expand_btn{display:none;}} .gvEcCq .floats-partial-section .floats-partial-footer .floats-partial-caption .caption-text{-webkit-flex:9;-ms-flex:9;flex:9;text-align:left;padding:21px 37px 10px 0px;font-size:16px;font-weight:700;} @media (min-width:768px){.gvEcCq .floats-partial-section .floats-partial-footer .floats-partial-caption .caption-text{padding:19px 18px 10px 18px;}} @media (min-width:992px){.gvEcCq .floats-partial-section .floats-partial-footer .floats-partial-caption .caption-text{padding:19px 21px 10px 21px;}} .gvEcCq .floats-partial-section .floats-partial-footer .floats-partial-caption .caption-partial-url{-webkit-flex:1;-ms-flex:1;flex:1;background:url('data:image/svg+xml;base64,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') no-repeat 6px;background-position:center;padding:16px 0 16px 0;min-width:62px;max-width:62px;} @media (min-width:768px){.gvEcCq .floats-partial-section .floats-partial-footer .floats-partial-caption .caption-partial-url{padding:16px 0 17px 0;min-width:52px;max-width:52px;}} @media (min-width:992px){.gvEcCq .floats-partial-section .floats-partial-footer .floats-partial-caption .caption-partial-url{padding:15px 0 17px 0;min-width:60px;max-width:60px;}} .gvEcCq .floats-partial-section .floats-partial-footer .floats-partial-caption .caption-partial-url:hover{-webkit-text-decoration:none;text-decoration:none;} .gvEcCq .floats-partial-section .floats-partial-footer .floats-partial-caption-text{font-size:14px;padding:0px 18px 17px 22px;} @media (min-width:768px){.gvEcCq .floats-partial-section .floats-partial-footer .floats-partial-caption-text{padding:0px 15px 15px 18px;}} @media (min-width:992px){.gvEcCq .floats-partial-section .floats-partial-footer .floats-partial-caption-text{padding:0px 18px 19px 21px;}} .gvEcCq .floats-partial-section .floats-partial-footer .floats-partial-caption-text figcaption{margin:0;padding:0;} .gvEcCq .floats-partial-section .floats-partial-footer .floats-partial-caption-text svg{max-width:100%;} .gvEcCq ol{font-size:16px;padding-left:32px;font-family:STIXGeneral-Regular;line-height:24px;margin:15px 0;} .gvEcCq ol li{margin:10px 0;padding-left:10px;word-break:normal;} .gvEcCq ol li a{word-break:break-all;} .gvEcCq ol li .reflinks{display:block;} .gvEcCq ol li .reflinks a{color:#4D8A17;-webkit-text-decoration:none;text-decoration:none;cursor:pointer;} .gvEcCq ol li .reflinks a:hover{-webkit-text-decoration:underline;text-decoration:underline;} .gvEcCq ol li .reflinks .sep{color:#000000;} .gvEcCq .table-wrapper{width:auto;overflow-x:auto;} .gvEcCq .table-wrapper::-webkit-scrollbar{-webkit-appearance:none;width:5px;height:5px;} .gvEcCq .table-wrapper::-webkit-scrollbar-thumb{border-radius:5px;background-color:rgba(0,0,0,.5);-webkit-box-shadow:0 0 1px rgba(255,255,255,.5);} @media (min-width:768px){.gvEcCq .table-wrapper{overflow-x:auto;width:auto;-webkit-break-word:word-break;break-word:word-break;margin:0 auto;}} @media (min-width:992px){.gvEcCq .table-wrapper{width:auto;}} @media (min-width:1324px){.gvEcCq .table-wrapper{width:auto;}} @media (min-width:320px){.gvEcCq{margin-bottom:33px;padding:0;}} @media (min-width:992px){.gvEcCq{padding:0;margin-bottom:100px;}} @media (min-width:1325px){.gvEcCq{margin-bottom:100px;}} .gvEcCq .owl-stage{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;} .gvEcCq .floats-partial-middle--thumbs{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;} .gvEcCq .floats-partial-middle--thumbs .partial-carousel-dot svg{fill:#4D8A17;} .gvEcCq .floats-partial-middle--thumbs .partial-carousel-dot strong{color:#4D8A17;font-weight:500;} .gvEcCq .floats-partial-middle--thumbs .partial-carousel-dot.active strong{color:#007B8B;} .gvEcCq .floats-partial-middle--thumbs .partial-carousel-dot.active svg{fill:#007B8B;} .gvEcCq .floats-partial-middle--thumbs .partial-carousel-dot .nowrap{white-space:nowrap;} .gvEcCq .partial-carousel-dot{text-align:center;cursor:pointer;border:0;width:100px;padding-bottom:20px;background-color:#fff;display:table-cell;} .gvEcCq .partial-carousel-dot:focus-visible{outline-offset:-3px;} .gvEcCq .partial-carousel-dot strong{font-size:11px;padding-top:12px;} @media (min-width:767.5px){.gvEcCq .partial-carousel-dot strong{font-size:16px;}} .gvEcCq .partial-carousel-dot img{width:90px;height:90px;border:1px solid #979797;margin:10px 0;} .gvEcCq .partial-carousel-wrapper .floats-partial-middle{border-top:1px solid #e3e3e3;display:block;margin-bottom:50px;padding-top:1px;} @media (min-width:767.5px){.gvEcCq .partial-carousel-wrapper .floats-partial-middle{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;margin-bottom:0;}} .gvEcCq .partial-carousel--single-image{overflow:hidden;text-align:center;} .gvEcCq .partial-carousel--single-image img{width:100%;max-height:239px;height:auto;} @media (min-width:767.5px){.gvEcCq .partial-carousel--single-image img{max-height:357px;}} @media (min-width:991.5px){.gvEcCq .partial-carousel--single-image img{max-height:417px;}} .gvEcCq .partial-carousel--single-image strong{font-size:11px;padding-top:12px;} @media (min-width:767.5px){.gvEcCq .partial-carousel--single-image strong{font-size:16px;}} .gvEcCq .owl-nav,.gvEcCq .owl-dots{display:none;} .gvEcCq .partial-carousel-wrapper .owl-loaded{position:relative;height:auto;max-height:287px;} @media (min-width:767.6px){.gvEcCq .partial-carousel-wrapper .owl-loaded{max-height:414px;}} @media (min-width:767.5px){.gvEcCq .partial-carousel-wrapper .owl-loaded{max-height:477px;}} .gvEcCq .partial-carousel-wrapper .floats-partial-content,.gvEcCq .owl-stage-outer,.gvEcCq .owl-stage{max-height:287px;height:auto;} @media (min-width:767.6px){.gvEcCq .partial-carousel-wrapper .floats-partial-content,.gvEcCq .owl-stage-outer,.gvEcCq .owl-stage{max-height:414px;}} @media (min-width:767.5px){.gvEcCq .partial-carousel-wrapper .floats-partial-content,.gvEcCq .owl-stage-outer,.gvEcCq .owl-stage{max-height:477px;}} .gvEcCq .partial-carousel-wrapper .floats-partial-content::after,.gvEcCq .owl-stage-outer::after,.gvEcCq .owl-stage::after{background:transparent;} .gvEcCq .partial-carousel-wrapper .floats-partial-middle--thumbs{width:100%;overflow-x:auto;-webkit-align-items:self-start;-webkit-box-align:self-start;-ms-flex-align:self-start;align-items:self-start;} @media (min-width:767.5px){.gvEcCq .partial-carousel-wrapper .floats-partial-middle--thumbs{width:calc(100% - 110px);}} .gvEcCq .partial-carousel-wrapper .floats-partial-middle--nav{width:100%;position:relative;} @media (min-width:767.5px){.gvEcCq .partial-carousel-wrapper .floats-partial-middle--nav{width:20%;}} .gvEcCq .floats-partial-middle{position:relative;} .gvEcCq .partial-carousel-wrapper .carousel-prev{background:url('data:image/svg+xml;base64,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') no-repeat 17.5px;height:50px;width:61px;display:inline-block;border:0;border-right:1px solid #E3E3E3;padding:0;cursor:pointer;background-position-y:54.35%;} @media (min-width:767.5px){.gvEcCq .partial-carousel-wrapper .carousel-prev{width:51px;background-position-x:8px;}} @media (min-width:991.5px){.gvEcCq .partial-carousel-wrapper .carousel-prev{width:59px;background-position-x:16px;}} .gvEcCq .partial-carousel-wrapper .carousel-next{background:url('data:image/svg+xml;base64,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') no-repeat 18.5px;height:52px;width:61px;display:inline-block;padding:0px;cursor:pointer;border:0;} @media (min-width:767.5px){.gvEcCq .partial-carousel-wrapper .carousel-next{width:51px;}} @media (min-width:991.5px){.gvEcCq .partial-carousel-wrapper .carousel-next{width:59px;}} .gvEcCq .partial-carousel-wrapper .partial-carousel-next-prev{position:absolute;right:0;bottom:-50px;height:50px;} @media (min-width:767.5px){.gvEcCq .partial-carousel-wrapper .partial-carousel-next-prev{bottom:0;}} .gvEcCq .partial-carousel-wrapper .floats-partial-content{padding:12px;} .gvEcCq .partial-carousel-wrapper .floats-partial-section .floats-partial-footer .floats-partial-caption .caption-partial-url{border-left:none;} .gvEcCq .partial-other-wrapper[data-section*="tab"] .floats-partial-content{display:none;} .gvEcCq .partial-other-wrapper[data-section*="tab"] .floats-partial-footer{background-image:url(https://w2o-xhtml-parsed-preprod.s3-eu-west-1.amazonaws.com/table_thumb.svg);background-repeat:no-repeat;background-position:17px 17px;background-size:100px 70px;border-top:none;min-height:110px;position:relative;} .gvEcCq .partial-other-wrapper[data-section*="tab"] .floats-partial-footer .table-thumb-overlay{cursor:pointer;position:absolute;left:17px;top:17px;height:70px;width:100px;} .gvEcCq .partial-other-wrapper[data-section*="tab"] .floats-partial-footer .caption-text{padding-left:137px;padding-right:0px;position:relative;} .gvEcCq .partial-other-wrapper[data-section*="tab"] .floats-partial-footer .caption-text .caption-text-overlay{cursor:pointer;position:absolute;top:20px;left:135px;height:20px;width:60px;} .gvEcCq .partial-other-wrapper[data-section*="tab"] .floats-partial-footer .floats-partial-caption-text{color:black;display:block !important;-webkit-text-decoration:none;text-decoration:none;width:auto;margin-left:140px;margin-right:20px;padding:0px 0px 15px 0px;width:-webkit-fit-content;width:-moz-fit-content;width:fit-content;} .gvEcCq .partial-other-wrapper[data-section*="tab"] .floats-partial-footer .minimize_btn{display:none !important;} .gvEcCq .partial-other-wrapper[data-section*="tab"] .floats-partial-footer .expand_btn{display:none !important;} @media (max-width:567px){.gvEcCq .partial-other-wrapper[data-section*="tab"] .floats-partial-footer{background-size:60px 45px;}.gvEcCq .partial-other-wrapper[data-section*="tab"] .floats-partial-footer .table-thumb-overlay{height:45px;width:60px;}.gvEcCq .partial-other-wrapper[data-section*="tab"] .floats-partial-footer .caption-text{padding:30px 0px 15px 95px;}.gvEcCq .partial-other-wrapper[data-section*="tab"] .floats-partial-footer .caption-text .caption-text-overlay{top:30px;left:95px;}.gvEcCq .partial-other-wrapper[data-section*="tab"] .floats-partial-footer .floats-partial-caption-text{margin:0px 20px;}} .gvEcCq .owl-item{overflow-y:hidden;margin-bottom:0;display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-align-items:center;-webkit-box-align:center;-ms-flex-align:center;align-items:center;-webkit-box-pack:center;-webkit-justify-content:center;-ms-flex-pack:center;justify-content:center;} .gvEcCq .supplementary--docs{display:-ms-flexbox;display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-ms-flex-wrap:wrap;-webkit-flex-wrap:wrap;-ms-flex-wrap:wrap;flex-wrap:wrap;padding-left:2em;} @media (min-width:767.5px){.gvEcCq .supplementary--docs{padding-left:1em;}} .gvEcCq .supplementary--docs li{-ms-flex:0 0 100%;-webkit-flex:0 0 100%;-ms-flex:0 0 100%;flex:0 0 100%;max-width:100%;} @media (min-width:991.5px){.gvEcCq .supplementary--docs li{-ms-flex:0 0 50%;-webkit-flex:0 0 50%;-ms-flex:0 0 50%;flex:0 0 50%;max-width:50%;}} .gvEcCq .supplementary--docs li a::after{content:'';background :url(data:image/svg+xml;base64,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) no-repeat 5px;padding-right:35px;background-size:14px;} .gvEcCq .dataset-section .supplementary--dpis-desc{padding-left:19px;} .gvEcCq .dataset-section .supplementary--dpis-desc p{margin:10px 0;} .gvEcCq .table-group.fixed-width,.gvEcCq .table-group.sidewaystable{min-width:auto;} .gvEcCq .noliststyle{padding-left:1em;} .gvEcCq .noliststyle li{padding-left:0;} .gvEcCq .xml-content > hr{display:none;} .gvEcCq sup,.gvEcCq sub{vertical-align:baseline;position:relative;top:5px;} .gvEcCq sup{top:-5px;} @media (min-width:767.5px){.gvEcCq a{position:relative;}} .gvEcCq a div.tooltip{cursor:auto;position:absolute;width:280px;background:#fff;border:1px solid #adadad;padding:20px;margin:0;display:none;z-index:2;color:#000;} @media (min-width:767.5px){.gvEcCq a div.tooltip{width:400px;}} @media (min-width:991.5px){.gvEcCq a div.tooltip{width:572px;}} .gvEcCq a div.tooltip .cross{position:absolute;right:10px;top:10px;display:block;height:14px;width:14px;cursor:pointer;} .gvEcCq a div.tooltip ol{margin:0;padding:0 0 10px 20px;} .gvEcCq a div.tooltip ol li{text-align:left;margin:0;} .gvEcCq a div.tooltip.left{left:20px;} @media (min-width:767.5px){.gvEcCq a div.tooltip.left{left:-20px;}} @media (min-width:991.5px){.gvEcCq a div.tooltip.left{left:-100px;}} .gvEcCq a div.tooltip.right{right:20px;} @media (min-width:767.5px){.gvEcCq a div.tooltip.right{right:-20px;}} @media (min-width:991.5px){.gvEcCq a div.tooltip.right{right:-100px;}} .gvEcCq a div.tooltip a.seeRef{padding-left:30px;} .gvEcCq .gloss-abbr tr td:first-child{white-space:nowrap;} .gvEcCq table.algorithm td{word-break:keep-all;} .gvEcCq h4#copyright + p a{word-break:break-word;} /* sc-component-id: sc-bGbJRg */ .dVCLgr{width:100%;} @media (min-width:991.5px){.dVCLgr{width:19%;}} @media (min-width:1280px){.dVCLgr{width:24%;max-width:350px;}} .dVCLgr p{border:1px solid #E3E3E3;} /* sc-component-id: sc-bEjcJn */ .jLUVhb{-webkit-flex:1 0 30%;-ms-flex:1 0 30%;flex:1 0 30%;list-style:none;padding:0;-webkit-transition:-webkit-transform 0.5s;-webkit-transition:transform 0.5s;transition:transform 0.5s;} @media (min-width:320px) and (max-width:991.5px){.jLUVhb{position:fixed;width:100%;z-index:2;margin:0;margin-top:50px;top:169px;max-height:calc(100vh - 165px);overflow-y:auto;}} @media (min-width:767.5px) and (max-width:991.5px){.jLUVhb{top:179px;}} @media (min-width:991.5px) and (max-width:1200px){.jLUVhb{max-width:354px;-webkit-flex:1 0 20%;-ms-flex:1 0 20%;flex:1 0 20%;}} @media (min-width:1200px){.jLUVhb{-webkit-flex:1 0 25%;-ms-flex:1 0 25%;flex:1 0 25%;max-width:354px;}} .jLUVhb p{margin:0;padding:10px 0;font-size:16px;font-weight:700;line-height:24px;display:none;} @media (min-width:991.5px){.jLUVhb p{display:block;padding:10px 20px;border:1px solid #E3E3E3;border-bottom:0;}} .jLUVhb.links__hidden{display:none;} @media (min-width:991.5px){.jLUVhb.links__hidden{display:block;}} .jLUVhb.translate{-webkit-transform :translateY(-120px);-ms-transform :translateY(-120px);transform :translateY(-120px);} @media (min-width:768px){.jLUVhb.translate{-webkit-transform:translateY(-130px);-ms-transform:translateY(-130px);transform:translateY(-130px);}} @media (min-width:991.5px){.jLUVhb.translate{-webkit-transform:none;-ms-transform:none;transform:none;}.jLUVhb.translate #leftSectionMenu{-webkit-transform:translateY(-130px);-ms-transform:translateY(-130px);transform:translateY(-130px);}} /* sc-component-id: sc-ePZHVD */ .ewHMLz{color:#4A4A4A;} /* sc-component-id: sc-likbZx */ .bBmZJZ{font-size:16px;font-weight:500;line-height:20px;color:#4A4A4A;border:1px solid #E3E3E3;border-bottom:0;background-color:#f2f2f2;display:list-item;} @media (min-width:991px){.bBmZJZ{-webkit-transition:background-color 0.25s ease-in-out 0s;transition:background-color 0.25s ease-in-out 0s;}.bBmZJZ:hover{background-color :#D8D8D8;-webkit-transition:background-color 0.25s ease-in-out 0s;transition:background-color 0.25s ease-in-out 0s;}} .bBmZJZ:last-child{border-bottom:1px solid #E3E3E3;} .bBmZJZ a{display:block;padding:10px 20px;} /* sc-component-id: sc-eKZiaR */ .dzZrgS{position:fixed;width:100%;height:50px;color:#4D8A17;background-color:#F2F2F2;border:none;font-size:16px;line-height:18px;font-weight:500;z-index:2;top:170px;-webkit-transition:-webkit-transform 0.5s;-webkit-transition:transform 0.5s;transition:transform 0.5s;text-align:left;border-top:1px solid #FFFFFF;} @media (min-width:320px) and (max-width:991.5px){.dzZrgS{padding-left:15px;border-bottom:4px solid #FFFFFF;}.dzZrgS.is-open{border-bottom:4px solid #F2F2F2;}} @media (min-width:767.5px){.dzZrgS{top:179px;}} @media (min-width:991.5px){.dzZrgS{display:none;}} .dzZrgS .icon__dots{cursor:pointer;-webkit-transform:rotate(90deg) translateY(8px);-ms-transform:rotate(90deg) translateY(8px);transform:rotate(90deg) translateY(8px);margin-right:15px;} @media (min-width:991.5px){.dzZrgS .icon__dots{padding-right:20px;margin-bottom:2px;-webkit-transform:none;-ms-transform:none;transform:none;}} .dzZrgS.translate{-webkit-transform :translateY(-120px);-ms-transform :translateY(-120px);transform :translateY(-120px);top:170px;} @media(min-width:991.5px){.dzZrgS.translate{-webkit-transform :translateY(-130px) rotate(90deg);-ms-transform :translateY(-130px) rotate(90deg);transform :translateY(-130px) rotate(90deg);}} /* sc-component-id: sc-jrIrqw */ .eHMFfo{padding:0;margin:20px 0;} .eHMFfo span{font-size:14px;} .eHMFfo span.isHide{display:none;} .eHMFfo span i{font-style:italic;} .eHMFfo span i.isHide{display:none;} /* sc-component-id: sc-hjRWVT */ .irNWJI{width:16px;margin:0 2px 2px;} /* sc-component-id: sc-iybRtq */ .iQSObO{width:16px;margin-bottom:2px;} /* sc-component-id: sc-iQtOjA */ .kfDwZY{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;border:none;font-size:12px;font-weight:500;margin-top:7px;padding:0;color:#4D8A17;-webkit-text-decoration:none;text-decoration:none;cursor:pointer;} .kfDwZY:hover{-webkit-text-decoration:underline;text-decoration:underline;} @media (min-width:767.5px){.kfDwZY{margin-top:5px;}} /* sc-component-id: sc-fHxwqH */ .dMuGGA{overflow-y:hidden;height:auto;max-height:100%;-webkit-transition:all .5s;transition:all .5s;} .dMuGGA.isHide{max-height:0;-webkit-transition:all .5s;transition:all .5s;} .dMuGGA p:first-child{margin-top:15px;} .dMuGGA p{font-size:14px;line-height:23px;display:block;} .dMuGGA sup{margin-top:-5px;} /* sc-component-id: sc-cEvuZC */ .XkrFe{border:none;border-top:1px solid #E3E3E3;height:2px;margin:15px 0;} .XkrFe.hr__author{margin-bottom:12px;} /* sc-component-id: sc-cBdUnI */ @media (min-width:320px) and (max-width:991.5px){.hRnXob{display:none;}} @media (min-width:991.5px){.hRnXob{z-index:1;}} /* sc-component-id: sc-exkUMo */ .hymroZ{background-color:#6ba339;color:#ffffff;display:none;} .hymroZ span.text{padding-left:20px;} @media (min-width:991px){.hymroZ{-webkit-transition:background-color 0.25s ease-in-out 0s;transition:background-color 0.25s ease-in-out 0s;}.hymroZ:hover{background-color :#D8D8D8;-webkit-transition:background-color 0.25s ease-in-out 0s;transition:background-color 0.25s ease-in-out 0s;}} /* sc-component-id: sc-kcDeIU */ .cbcyj{background-color:#f2f2f2;color:#4a4a4a;display:none;} .cbcyj span.text{padding-left:20px;} @media (min-width:991.5px) and (max-width:1279.5px){.cbcyj{display:none !important;}} @media (min-width:991px){.cbcyj{-webkit-transition:background-color 0.25s ease-in-out 0s;transition:background-color 0.25s ease-in-out 0s;}.cbcyj:hover{background-color :#D8D8D8;-webkit-transition:background-color 0.25s ease-in-out 0s;transition:background-color 0.25s ease-in-out 0s;}} @media (min-width:991.5px){.cbcyj.hide{display:none !important;}} /* sc-component-id: sc-BngTV */ .hmDWcT{background-color:#f2f2f2;color:#4a4a4a;display:none;} .hmDWcT span.text{padding-left:20px;} @media (min-width:1279.5px){.hmDWcT{display:none !important;}} @media (min-width:991px){.hmDWcT{-webkit-transition:background-color 0.25s ease-in-out 0s;transition:background-color 0.25s ease-in-out 0s;}.hmDWcT:hover{background-color :#5D9230;-webkit-transition:background-color 0.25s ease-in-out 0s;transition:background-color 0.25s ease-in-out 0s;}} @media (min-width:991.5px){.hmDWcT.hide{display:none !important;}} /* sc-component-id: sc-hBbWxd */ .kdRRST{background-color:#f2f2f2;border-top:1px solid #ffffff;display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;font-size:14px;-webkit-box-pack:justify;-webkit-justify-content:space-between;-ms-flex-pack:justify;justify-content:space-between;padding:20px 55px 20px 25px;} @media (max-width:1279.5px){.kdRRST{padding-right:20px;}} @media (max-width:991.5px){.kdRRST{display:none;}} /* sc-component-id: sc-dyGzUR */ .cHatnS div{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-flex-direction:column;-ms-flex-direction:column;flex-direction:column;} .cHatnS div:last-of-type{padding-top:20px;} .cHatnS span{color:#4F4F4F;font-size:20px;font-weight:700;line-height:26px;} /* sc-component-id: sc-drKuOJ */ .kxdGAZ{-webkit-align-items:center;-webkit-box-align:center;-ms-flex-align:center;align-items:center;display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-flex-direction:column-reverse;-ms-flex-direction:column-reverse;flex-direction:column-reverse;height:100%;-webkit-box-pack:end;-webkit-justify-content:flex-end;-ms-flex-pack:end;justify-content:flex-end;} .kxdGAZ span{padding-top:15px;} /* sc-component-id: sc-cTjmhe */ @media (min-width:991.5px){.ihmyEq{width:100%;z-index:1;}} /* sc-component-id: sc-cugefK */ .itFKcQ{background-color:#F2F2F2;color:#4A4A4A;border-top:1px solid #FFFFFF;font-size:16px;font-weight:500;padding:20px;display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-align-items:center;-webkit-box-align:center;-ms-flex-align:center;align-items:center;cursor:pointer;-webkit-box-pack:justify;-webkit-justify-content:space-between;-ms-flex-pack:justify;justify-content:space-between;} @media (min-width:991px){.itFKcQ{-webkit-transition:background-color 0.25s ease-in-out 0s;transition:background-color 0.25s ease-in-out 0s;}.itFKcQ:hover{background-color :#D8D8D8;-webkit-transition:background-color 0.25s ease-in-out 0s;transition:background-color 0.25s ease-in-out 0s;}} @media (min-width:991.5px){.itFKcQ.hide{display:none !important;}} .itFKcQ.active .icon-themeable{-webkit-transform:rotate(180deg);-ms-transform:rotate(180deg);transform:rotate(180deg);} /* sc-component-id: sc-fnwBNb */ .etzuBr{display:none;} /* sc-component-id: sc-iNhVCk */ .gWgCNb{background-color:#f2f2f2;border-radius:0;border-top:1px solid #e3e3e3;font-size:16px;color:#4a4a4a;font-weight:500;padding:11px 20px;display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-align-items:center;-webkit-box-align:center;-ms-flex-align:center;align-items:center;} @media (min-width:991px){.gWgCNb{-webkit-transition:background-color 0.25s ease-in-out 0s;transition:background-color 0.25s ease-in-out 0s;}.gWgCNb:hover{background-color :#D8D8D8;-webkit-transition:background-color 0.25s ease-in-out 0s;transition:background-color 0.25s ease-in-out 0s;}} /* sc-component-id: sc-eAKXzc */ .WHkfb{-webkit-flex:1 0 90%;-ms-flex:1 0 90%;flex:1 0 90%;padding-left:50px;} @media (min-width:767.5px){.WHkfb{padding-left:93px;}} @media (min-width:991.5px){.WHkfb{padding-left:50px;}} /* sc-component-id: sc-bfYoXt */ .cGuwDn{width:20px;height:24px;} /* sc-component-id: sc-gbOuXE */ .jGPvPL{width:22px;height:21px;padding:2px;} /* sc-component-id: sc-dRFtgE */ .jrsDZP{-webkit-flex:1 0 80%;-ms-flex:1 0 80%;flex:1 0 80%;padding-left:26px;} @media (min-width:767.5px){.jrsDZP{padding-left:69px;}} @media (min-width:991.5px){.jrsDZP{padding-left:20px;}} @media (min-width:991.5px) and (max-width:1279.5px){.jrsDZP{display:none;}} /* sc-component-id: sc-gkFcWv */ .gWTCXQ{display:none;} @media (min-width:991.5px){.gWTCXQ{padding-left:20px;max-width:100px;display:inline;}} @media (min-width:1279.5px){.gWTCXQ{display:none;}} /* sc-component-id: sc-hUfwpO */ .jHPgUP{width:100%;border-left:1px solid #f2f2f2;border-right:1px solid #f2f2f2;border-bottom:none;cursor:pointer;} .jHPgUP li{background-color:#f2f2f2;} @media (min-width:991px){.jHPgUP li{-webkit-transition:background-color 0.25s ease-in-out 0s;transition:background-color 0.25s ease-in-out 0s;}.jHPgUP li:hover{background-color :#ffffff;-webkit-transition:background-color 0.25s ease-in-out 0s;transition:background-color 0.25s ease-in-out 0s;}} .jHPgUP h3{display:none;} /* sc-component-id: sc-imABML */ .ibWUzs{background-color:#f2f2f2;color:#4a4a4a;display:block;border-top:1px solid #ffffff;border-radius:0;font-size:16px;font-weight:500;padding:20px;display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-align-items:center;-webkit-box-align:center;-ms-flex-align:center;align-items:center;cursor:pointer;-webkit-box-pack:justify;-webkit-justify-content:space-between;-ms-flex-pack:justify;justify-content:space-between;} .ibWUzs .right__icon{padding:0;height:16px;width:20px;} @media (min-width:991px){.ibWUzs{-webkit-transition:background-color 0.25s ease-in-out 0s;transition:background-color 0.25s ease-in-out 0s;}.ibWUzs:hover{background-color :#D8D8D8;-webkit-transition:background-color 0.25s ease-in-out 0s;transition:background-color 0.25s ease-in-out 0s;}} @media (min-width:991.5px){.ibWUzs.hide{display:none !important;}} /* sc-component-id: sc-cjHlYL */ .hYnXNL{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;width:25%;float:right;margin-top:-49px;} @media (max-width:374px){.hYnXNL{width:26.5%;}} @media (min-width:992px){.hYnXNL{display:none;}} /* sc-component-id: sc-dRaagA */ .hGCBfz{background-color:#6BA339;color:white;-webkit-flex:1 0 25%;-ms-flex:1 0 25%;flex:1 0 25%;z-index:100;padding:14px 16px 10px 19px;border-radius:0px;border-bottom:1px solid #FFFFFF;} @media (min-width:320px) and (max-width:767px){.hGCBfz{border-bottom:none;padding:16px 16px 14px 16px;font-size:14px;}.hGCBfz .article_toolbar_icon{display:none;}} /* sc-component-id: sc-fEUNkw */ .dfaOeH{margin-right:16px;width:24px;height:24px;} /* sc-component-id: sc-dREXXX */ .cqhPZs{width:16px;height:auto;float:right;} @media (min-width:767.5px){.cqhPZs{width:24px;height:24px;}} /* sc-component-id: sc-nrwXf */ .byflLi{font-size:14px;padding:16px 20px 24px;} @media (max-width:991px){.byflLi{display:none;}} /* sc-component-id: sc-eitiEO */ .jXeALb{font-size:14px;} .jXeALb:hover{-webkit-text-decoration:underline;text-decoration:underline;} /* sc-component-id: sc-bhlBdH */ .UVThf{font-size:14px;font-weight:bold;} /* sc-component-id: sc-OxbzP */ .eEFYIL{width:100%;height:auto;background-color:#4A4A4A;font-family:'IBM Plex Sans',sans-serif;font-size:14px;font-weight:500;color:#fff;padding:15px 20px;line-height:20px;position:fixed;bottom:0px;z-index:3;display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-box-pack:justify;-webkit-justify-content:space-between;-ms-flex-pack:justify;justify-content:space-between;-webkit-transition:-webkit-transform .25s;-webkit-transition:transform .25s;transition:transform .25s;} .eEFYIL a{color:#7ED321;} .eEFYIL p{display:inline;margin:0;} .eEFYIL.hideBanner{-webkit-transform:translateY(100%);-ms-transform:translateY(100%);transform:translateY(100%);} /* sc-component-id: sc-bYnzgO */ .dsXEHZ{background:transparent;border:none;height:20px;width:20px;margin-left:20px;padding:0;cursor:pointer;} /* sc-component-id: sc-cPuPxo */ .iXKuYd{min-width:20px;height:20px;} /* sc-component-id: baseContent__Main-sc-134g3di-2 */ .cmzomn{margin-top:140px;} @media (min-width:767.5px){.cmzomn{margin-top:85px;}} @media (min-width:991.5px){.cmzomn{margin-top:70px;min-height:calc(100vh - 334px);}} /* sc-component-id: threeColumn__ThreeColumnWrapper-sc-1bf4078-0 */ .dDyxFF{width:100%;} @media only screen and (min-width:992px){.dDyxFF{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-flex-flow:row wrap;-ms-flex-flow:row wrap;flex-flow:row wrap;padding:0 20px;max-width:1440px;margin:0 auto;}} @media (min-width:991.5px) and (max-width:1100.5px){.dDyxFF .partenership__collabText{font-size:12px;}} @media (min-width:1100.5px) and (max-width:1200.5px){.dDyxFF .partenership__collabText{font-size:14px;}} @media (min-width:991.5px) and (max-width:1071.5px){.dDyxFF .partenership__logoWrapper img{margin-left:20px;margin-bottom:10px;}.dDyxFF .partenership__logoWrapper img:last-child{margin-left:0;margin-bottom:0;}} @media (max-width:1150px){.dDyxFF .journal-metrics-container{display:none;}.dDyxFF .journal_report_link-border-helper{display:none;}} /* sc-component-id: threeColumn__LeftWrapper-sc-1bf4078-1 */ @media only screen and (min-width:992px){.fkVFlF{width:20%;}} @media only screen and (min-width:1280px){.fkVFlF{width:25%;}} /* sc-component-id: threeColumn__RightWrapper-sc-1bf4078-2 */ .jzTMLK{-webkit-transition:-webkit-transform 0.5s;-webkit-transition:transform 0.5s;transition:transform 0.5s;} @media only screen and (min-width:992px){.jzTMLK{width:20%;margin-bottom:60px;}.jzTMLK.translate{-webkit-transform:translateY(-130px);-ms-transform:translateY(-130px);transform:translateY(-130px);}} @media only screen and (min-width:1280px){.jzTMLK{width:25%;}} /* sc-component-id: threeColumn__ContentWrapper-sc-1bf4078-3 */ .bbzREB{margin:0 20px;overflow:hidden;} @media only screen and (min-width:992px){.bbzREB{width:calc(60% - 40px);}} @media only screen and (min-width:1280px){.bbzREB{width:calc(50% - 40px);}} /* sc-component-id: threeColumn__Content-sc-1bf4078-4 */ .jknsZc{margin:0 auto;overflow:hidden;} @media only screen and (min-width:768px){.jknsZc{max-width:688px;}} @media only screen and (min-width:992px){.jknsZc{max-width:100%;}} @media only screen and (min-width:1075px){.jknsZc{max-width:572px;}} /* sc-component-id: article__BaseStyle-sc-10qhnp8-0 */ .gnduKR{width:100%;margin-top:170px;} @media (min-width:767.5px){.gnduKR{margin-top:179px;}} @media (min-width:991.5px){.gnduKR{margin-top:180px;min-height:840px;}} /* sc-component-id: article__StickyStyle-sc-10qhnp8-1 */ .fJLQxX{width:100%;max-width:1440px;z-index:3;-webkit-transition:all 0.5s;transition:all 0.5s;} .fJLQxX.translate{-webkit-transform:translateY(-120px);-ms-transform:translateY(-120px);transform:translateY(-120px);} @media (min-width:767.5px){.fJLQxX.translate{-webkit-transform:translateY(-130px);-ms-transform:translateY(-130px);transform:translateY(-130px);}} @media (min-width:991.5px){.fJLQxX.translate{-webkit-transform:translateY(-130px);-ms-transform:translateY(-130px);transform:translateY(-130px);}} /* sc-component-id: article__SubtitleStyled-sc-10qhnp8-2 */ .elhyhe{background-color:#007b8b;opacity:0.8;color:#fff;margin-bottom:0;margin-top:0;padding:17px 0 17px 20px;line-height:23px;height:60px;-webkit-align-items:center;-webkit-box-align:center;-ms-flex-align:center;align-items:center;display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;font-size:20px;font-weight:500;} @media (min-width:767.5px){.elhyhe{line-height:26px;}} @media (min-width:991.5px){.elhyhe{line-height:50px;padding:5px 0 5px 20px;height:60px;}} /* sc-component-id: article__JournalMenuTitle-sc-10qhnp8-3 */ .jCfZqq{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-align-items:center;-webkit-box-align:center;-ms-flex-align:center;align-items:center;font-size:16px;font-weight:500;line-height:20px;height:50px;width:100%;background-color:#007b8b;color:#fff;text-align:left;padding-left:14px;padding-right:0;border:0;border-top:1px solid #fff;} .jCfZqq span{font-size:24px;padding-right:14px;} @media (min-width:991.5px){.jCfZqq{display:none;}} /* sc-component-id: article__WhiteWrapper-sc-10qhnp8-5 */ .bJcSpM{background-color:#fff;display:block;position:relative;} .bJcSpM h1,.bJcSpM h2{color:#fff;} .bJcSpM h2{font-size:20px;} @media (min-width:767.5px){.bJcSpM h2{font-size:25px;}} .bJcSpM h1{font-size:20px;} @media (min-width:767.5px){.bJcSpM h1{font-size:24px;}} @media (min-width:991.5px){.bJcSpM h1{font-size:40px;}} /* sc-component-id: articleContent__ArticleContentsWrapper-sc-2yl9jy-0 */ .eSvjPB{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;width:100%;} .eSvjPB .author__editor--hr{display:block;} /* sc-component-id: articleContent__ArticleContentsSectionsWrapper-sc-2yl9jy-1 */ .jntdRu{width:100%;} @media (max-width:991.5px){.jntdRu{margin-top:52px;}} .jntdRu p{margin:12px 0;line-height:20px;} .jntdRu .articlecontent__alignment{width:100%;min-width:306px;padding-left:7px;padding-right:7px;} @media (min-width:346px){.jntdRu .articlecontent__alignment{padding-left:20px;padding-right:20px;}} @media (min-width:767.5px){.jntdRu .articlecontent__alignment{width:543px;margin:auto;padding-left:0px;padding-right:0px;}} @media (min-width:991.5px){.jntdRu .articlecontent__alignment{max-width:612px;width:75%;margin:inherit;margin-left:50px;padding-left:20px;padding-right:20px;}} @media (min-width:1100px){.jntdRu .articlecontent__alignment{margin-left:118px;}} .jntdRu .article__title{padding-bottom:12px;} @media (min-width:767.5px){.jntdRu .article__title{padding-bottom:15px;}} .jntdRu .article__summary{padding-bottom:65px;} @media (min-width:991.5px){.jntdRu .article__summary{padding-bottom:75px;}} @media (min-width:767.5px){.jntdRu #vjs_video_3 button{-webkit-transform:none;-ms-transform:none;transform:none;left:40%;margin-top:0;}} .jntdRu .video-js .vjs-big-play-button{top:calc(50% - 22.5px);left:calc(50% - 50px);} .jntdRu .video-js{margin:20px 0;} .jntdRu a{color:#4D8A17;-webkit-text-decoration:none;text-decoration:none;cursor:pointer;} .jntdRu a:hover{-webkit-text-decoration:underline;text-decoration:underline;} /* sc-component-id: articleContent__ArticleWrapper-sc-2yl9jy-2 */ .hAzjHP{background-position:0px 0px;background-origin:content-box;background-size:99%;} /* sc-component-id: articleContent__EditorWrapper-sc-2yl9jy-3 */ .evMkvU{padding:0;} /* sc-component-id: articleContent__PublicationTimeLineWrapper-sc-2yl9jy-4 */ .fjviok{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-flex-wrap:wrap;-ms-flex-wrap:wrap;flex-wrap:wrap;margin:15px 0 30px;border-bottom:1px solid #E3E3E3;} @media (min-width:767.5px){.fjviok{border-top:1px solid #E3E3E3;margin:20px 0 30px;}} /* sc-component-id: articleContent__PublcationCol-sc-2yl9jy-5 */ .hgKieC{-webkit-flex-basis:0;-ms-flex-preferred-size:0;flex-basis:0;-webkit-box-flex:1;-webkit-flex-grow:1;-ms-flex-positive:1;flex-grow:1;max-width:100%;border-right:1px solid #E3E3E3;padding:5px 0 15px;} @media (min-width:767.5px){.hgKieC{padding:5px 0;}} .hgKieC.notPublished{display:none;} @media (min-width:767.5px){.hgKieC.notPublished{display:block;}} @media (max-width:767.5px){.hgKieC.isPublished{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;}.hgKieC.isPublished span:first-child::after{content:':';padding-right:3px;}} .hgKieC:last-child{border-right:none;} /* sc-component-id: articleContent__PublicationField-sc-2yl9jy-6 */ .cIribm{display:block;text-align:center;font-size:14px;font-weight:700;} /* sc-component-id: articleContent__PublicationValue-sc-2yl9jy-7 */ .khHUcu{display:block;text-align:center;font-size:14px;} /* sc-component-id: articleContent__TextWrapper-sc-2yl9jy-8 */ .fYrLtk{margin:-7px 0 -5px;font-size:14px;} /* sc-component-id: biblio__Wrapper-sc-1nou1lz-0 */ .eyHfWh{position:relative;} .eyHfWh.hide{display:none;} .eyHfWh hr{position:absolute;left:0;right:0;top:-30px;border-top:1px solid #E3E3E3;border-bottom:0;}.cBTaXT.hide{display:none;} .cBTaXT hr{display:none;} /* sc-component-id: biblio__BiblioWrapper-sc-1nou1lz-1 */ @font-face{font-family:'STIXGeneral-Bold';src:url('https://static-01.hindawi.com/next_assets/wyT4heYybLmt1UEJexPE9/_next/static/_/hindawi-ui/fonts/STIXGeneral-Regular.otf');} .hFGNjP{padding:0 20px 20px;} .hFGNjP .bib__image{display:none;} .hFGNjP .bib__title{font-size:20px;} .hFGNjP .bib__name{font-size:18px;font-family:'STIXGeneral-Bold',serif !important;color:rgb(74,74,74);font-weight:700;} .hFGNjP .bib--split .bib__link .bib__info,.hFGNjP .bib--split .bib__link.bib__link--image .bib__info,.hFGNjP [class*='bib--txt'] .bib__link .bib__info,.hFGNjP [class*='bib--txt'] .bib__link.bib__link--image .bib__info{padding:10px 0 !important;} .hFGNjP .bib__author{font-size:14px;line-height:22px;color:rgb(132,132,132);margin-top:3px;margin-bottom:0px;margin-right:20px;opacity:1;font-family:'IBM Plex Sans',sans-serif;font-weight:500;} .hFGNjP #biblio_sidebar .bib__link{border-top:1px solid rgb(227,227,227) !important;padding-top:5px !important;padding-bottom:5px !important;} .hFGNjP #biblio_sidebar .bib__link:first-child{border-top:0 !important;padding-top:0 !important;} .hFGNjP #biblio_sidebar .bib__link:first-child .bib__info{padding-top:0 !important;} .hFGNjP #biblio_widget .bib--split .bib__link .bib__info,.hFGNjP #biblio_widget .bib--split .bib__link.bib__link--image .bib__info{padding:10px 15px 10px 0 !important;} @media (min-width:767.5px){.hFGNjP{padding:0 40px 20px;}} @media (min-width:991.5px){.hFGNjP{padding:0 20px;}} /* sc-component-id: biblio__H4-sc-1nou1lz-2 */ .becpDt{font-size:20px;font-family:'IBM Plex Sans';color:#000;font-weight:600;margin-top:34px;margin-bottom:18px;padding:0 20px;} /* sc-component-id: journal_article__ArticleStickyTopToolbar-sc-12fk6zy-0 */ .gbrQxi{display:none;} @media (min-width:991.5px){.gbrQxi{display:block;}.gbrQxi .article_top_toolbar{top:110px;z-index:-100;}.gbrQxi .article_top_toolbar .pdf_desktop{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;}.gbrQxi .article_top_toolbar .ref_desktop{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;}} .gbrQxi .article_toolbar_views{margin-top:-58px;display:block;text-align:right;} /* sc-component-id: journal_article__ArticleStickyBottomToolbar-sc-12fk6zy-1 */ @media (min-width:991.5px){.isgRZA .article_bottom_toolbar{top:292px;z-index:-10;}} /* sc-component-id: journal_article__BreadCrumbStyle-sc-12fk6zy-2 */ @media (min-width:991.5px){.bjwAQR{padding:16px 20px 24px;height:60px;}} /* sc-component-id: journal_article__ArticleBaseStyle-sc-12fk6zy-3 */ .NQpXC .content,.NQpXC .contentWrapper{overflow:visible;} /* sc-component-id: journal_article__BiblioIntegratorSidebar-sc-12fk6zy-4 */ .cYsVdk{max-width:350px;} @media (max-width:991.5px){.cYsVdk{display:none !important;}} /* sc-component-id: journal_article__BiblioIntegratorWidget-sc-12fk6zy-5 */ .dbAANp{background-color:#f5f5f5;padding:15px 20px 24px 20px;margin-top:2px;margin-bottom:20px;} @media (min-width:991.5px){.dbAANp{margin-top:-64px;margin-bottom:40px;}} .dbAANp .hrTitleRef{padding:0;margin-top:0;font-family:'IBM Plex Sans',sans-serif;} .dbAANp #biblio_widget{padding:0;}</style> </head> <body style="margin:0;box-sizing:border-box"> <div id="__next"> <div class="static_page"> <div class="page no__focus__outline"> <header class="sc-fjdhpX jzXSTA"> <div class="sc-bZQynM bNmdOy E_HUmlU2Hj3PpkDmWzKPd header__padding"> <div class="sc-kAzzGY epRoim mobileLogoFlex _1fpqArXPCxgnlv2rXZC2sl _1SoDWGr_FVsil9klRfjnhb"> <a class="sc-htpNat bUhGXt link" href="//www.newsama.com/" aria-label="Home"><img alt="标志”年代rc=”http年代://preprod-admin.w2o.hindawi.com/themes/custom/hindawi_extended/logo.svg" class="sc-ckVGcZ gjctdt"></a> </div> <div class="sc-jzJRlG jGJLow mobileButtonFlex _2v1j-tLx3hx0tgKcpPzH6B _3wm8FgzqRuRWouJG31YgKx E_HUmlU2Hj3PpkDmWzKPd"> <nav class="sc-bxivhb ivbKJD"> <ul class="sc-ifAKCX fopjOz sc-kpOJdX kAzjOl"> <li class="navigation__link"><a class="sc-htpNat bUhGXt link link" href="//www.newsama.com/journals/" aria-label="Journals">期刊</一个></li> <li class="navigation__link"><a class="sc-htpNat bUhGXt link link" href="//www.newsama.com/publish-research/" aria-label="Publish with us">雷电竞下载苹果</a></li> <li class="navigation__link"><a class="sc-htpNat bUhGXt link link" href="//www.newsama.com/publishing-partnerships/" aria-label="Publishing partnerships">raybet推荐吗</a></li> <li class="navigation__link"><a class="sc-htpNat bUhGXt link link" href="//www.newsama.com/about/" aria-label="About us">关于我们</一个></li> <li class="navigation__link"><a class="sc-htpNat bUhGXt link link" href="//www.newsama.com/blog/" aria-label="Blog">博客</一个></li> </ul> </nav> <button class="sc-chPdSV hbVEoI icon__search" aria-label="Search"><span class="sc-htoDjs khbeDd icon__search icon-colored _3hSk3ArgsWMFh5QJIGblNr icon-themeable" title="" font-size="24"></span></button> <button aria-label="Hamburger Menu" class="sc-cSHVUG ldSkVQ"><span class="sc-htoDjs khbeDd hamburger icon-colored _20YUA8q5mREEBfJm8VsdXV icon-themeable" title="" font-size="24"></span></button> </div> </div> </header> <div class="sc-bZQynM bNmdOy"></div> <main class="baseContent__Main-sc-134g3di-2 cmzomn article__BaseStyle-sc-10qhnp8-0 gnduKR DADtd11SJ4KyWZ2l6zAnr base_style journal_article__ArticleBaseStyle-sc-12fk6zy-3 NQpXC"> <div class="sc-bZQynM bNmdOy width_wrapper"> <div id="journal__navigation" class="sc-cbkKFq kuxTps article__StickyStyle-sc-10qhnp8-1 fJLQxX"> <a class="sc-htpNat bUhGXt link article__WhiteWrapper-sc-10qhnp8-5 bJcSpM" href="//www.newsama.com/journals/am/" aria-label=""><h2 class="sc-eHgmQL fTkdCR article__SubtitleStyled-sc-10qhnp8-2 elhyhe _1ZrXcsbATVL_kHOpFAan9k">多媒体的发展</h2></一个><button class="article__JournalMenuTitle-sc-10qhnp8-3 jCfZqq journalNavCheck" id="journalMenuTitle"><span>+</年代p一个n>杂志菜单</button><div class="sc-bdVaJa EwsTm isHide" id="fadeOut"></div> <div class="sc-cjHlYL hYnXNL"> <a class="sc-htpNat bUhGXt link sc-dRaagA hGCBfz article-toolbar-link" href="http://downloads.hindawi.com/journals/am/2011/310762.pdf" aria-label="" target="_blank"><img alt="" class="sc-EHOje jOLhQl sc-fEUNkw dfaOeH article_toolbar_icon" title="" role="presentation" src="https://static-01.hindawi.com/next_assets/wyT4heYybLmt1UEJexPE9/_next/static/node_modules/hindawi-ui/src/icon/svg/pdf.svg" height="24">PDF<我米g alt="" class="sc-EHOje jOLhQl sc-dREXXX cqhPZs" title="" role="presentation" src="data:image/svg+xml;base64,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" height="24"></a> </div> <nav id="journal__nav" class="sc-hGoxap dEfJbW article__JournalNavigationStyle-sc-10qhnp8-4 gaRePw isHide journalNavCheck"> <div class="sc-giadOv eBepOs journalNavCheck"> <button class="sc-hSdWYo etauus sc-fjmCvl fSPaCp journalNavCheck" color="">日报》概述<年代p一个nclass="sc-htoDjs gvAGTV sc-bHwgHz ctePXh journalNavCheck icon__inactive icon-white _1CwL9R82TPl2HxDleeYGDR icon-themeable" title="" font-size="9"></span><span class="sc-htoDjs gvAGTV sc-bHwgHz ctePXh journalNavCheck icon__active icon-colored icon__hidden _1CwL9R82TPl2HxDleeYGDR icon-themeable" title="" font-size="9"></span></button> </div> <a class="sc-htpNat bUhGXt link sc-TFwJa gCIfTO" href="//www.newsama.com/journals/am/guidelines/" aria-label="For authors">对于作者</一个><一个class="sc-htpNat bUhGXt link sc-TFwJa gCIfTO" href="//www.newsama.com/publish-research/reviewers/" aria-label="For reviewers">对于审查员</一个><一个class="sc-htpNat bUhGXt link sc-TFwJa gCIfTO" href="//www.newsama.com/publish-research/editors/" aria-label="For editors">的编辑器</一个><一个class="sc-htpNat bUhGXt link sc-TFwJa gCIfTO" href="//www.newsama.com/journals/am/contents/year/2022/" aria-label="Table of Contents">表的内容</一个><div class="sc-giadOv eBepOs journalNavCheck"> <button class="sc-hSdWYo etauus sc-fjmCvl bVEEaH journalNavCheck" color="">特殊的问题<年代p一个nclass="sc-htoDjs gvAGTV sc-bHwgHz ctePXh journalNavCheck icon__inactive icon-white _1CwL9R82TPl2HxDleeYGDR icon-themeable" title="" font-size="9"></span><span class="sc-htoDjs gvAGTV sc-bHwgHz ctePXh journalNavCheck icon__active icon-colored icon__hidden _1CwL9R82TPl2HxDleeYGDR icon-themeable" title="" font-size="9"></span></button> </div> <a class="sc-htpNat bUhGXt link sc-jtRfpW hHKuAX submit__button journalNavCheck" href="https://review.hindawi.com/submit?journal=am" aria-label="" target="" id=""><span><img alt="" class="sc-EHOje jOLhQl sc-kTUwUJ jUTgTa" title="" role="presentation" src="data:image/svg+xml;base64,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" height="24"><span class="text">提交</年代p一个n></年代p一个n><span class="sc-htoDjs kNoiZu sc-dqBHgY cQgXqt fileArrow icon-colored undefined icon-themeable" title="" font-size="16"></span></a> <div class="sc-krDsej ewCAZX hide__submenu journalSubmenuCheck" id="journalnav__submenu"> <button class="sc-itybZL kdvWPs journalSubmenuCheck"><span class="sc-htoDjs gvAGTV sc-fzsDOv hAFFGk journalSubmenuCheck icon-colored _1CwL9R82TPl2HxDleeYGDR icon-themeable" title="" font-size="9"></span></button> </div> <div class="sc-dTdPqK lkWWFY hide__submenu journal__nav__desktop" id="journalnav__submenudesktop"> <div order="0" class="sc-kUaPvJ ehYbwJ"> <div class="sc-gPWkxV legory"></div> <div class="sc-gPWkxV legory"></div> </div> <div class="sc-jVODtj epgILe"></div> <span class="sc-fONwsr kCfRMJ"></span> </div> </nav> </div> <div class="sc-nrwXf byflLi breadcrumb__wrapper journal_article__BreadCrumbStyle-sc-12fk6zy-2 bjwAQR article"> <a class="sc-htpNat bUhGXt link sc-eitiEO jXeALb breadCrumb" href="//www.newsama.com/journals/am/" aria-label="Advances in Multimedia">多媒体的发展</一个>/<一个class="sc-htpNat bUhGXt link sc-eitiEO jXeALb breadCrumb" href="//www.newsama.com/journals/am/contents/year/2011/" aria-label="2011">2011年</一个>/<年代p一个nclass="sc-bhlBdH UVThf">文章</年代p一个n></div> <div class="threeColumn__ThreeColumnWrapper-sc-1bf4078-0 dDyxFF threeColumnWrapper"> <div class="threeColumn__LeftWrapper-sc-1bf4078-1 fkVFlF leftWrapper"> <button type="button" class="sc-eKZiaR dzZrgS section__button " id="sectionButton"><span class="sc-htoDjs diutvq icon__dots dots__detector icon-colored _3n-7Ho7DlfjMlZbrRr9Q4a icon-themeable" title="" font-size="22"></span>文章部分</button><div class="sc-bEjcJn jLUVhb static_contents__section_links links__hidden "> <div id="leftSectionMenu" class="sc-cbkKFq prcY sc-bGbJRg dVCLgr left_section_menu"> <p>在这一页上</p><年代p一个nclass="sc-likbZx bBmZJZ grey" property="background-color" color="#D8D8D8"><a class="sc-htpNat bUhGXt link sc-ePZHVD ewHMLz" href="#abstract" aria-label="">文摘</一个></年代p一个n><年代pan class="sc-likbZx bBmZJZ grey" property="background-color" color="#D8D8D8"><a class="sc-htpNat bUhGXt link sc-ePZHVD ewHMLz" href="#introduction" aria-label="">介绍</一个></年代p一个n><年代pan class="sc-likbZx bBmZJZ grey" property="background-color" color="#D8D8D8"><a class="sc-htpNat bUhGXt link sc-ePZHVD ewHMLz" href="#related-work" aria-label="">相关工作</一个></年代p一个n><年代pan class="sc-likbZx bBmZJZ grey" property="background-color" color="#D8D8D8"><a class="sc-htpNat bUhGXt link sc-ePZHVD ewHMLz" href="#experiments-and-evaluation" aria-label="">评价</一个></年代p一个n><年代pan class="sc-likbZx bBmZJZ grey" property="background-color" color="#D8D8D8"><a class="sc-htpNat bUhGXt link sc-ePZHVD ewHMLz" href="#conclusions" aria-label="">结论</一个></年代p一个n><年代pan class="sc-likbZx bBmZJZ grey" property="background-color" color="#D8D8D8"><a class="sc-htpNat bUhGXt link sc-ePZHVD ewHMLz" href="#references" aria-label="">引用</一个></年代p一个n><年代pan class="sc-likbZx bBmZJZ grey" property="background-color" color="#D8D8D8"><a class="sc-htpNat bUhGXt link sc-ePZHVD ewHMLz" href="#copyright" aria-label="">版权</一个></年代p一个n></div> </div> </div> <div class="threeColumn__ContentWrapper-sc-1bf4078-3 bbzREB contentWrapper"> <div class="threeColumn__Content-sc-1bf4078-4 jknsZc content"> <div class="articleContent__ArticleContentsWrapper-sc-2yl9jy-0 eSvjPB article_contents"> <div class="articleContent__ArticleContentsSectionsWrapper-sc-2yl9jy-1 jntdRu article_contents__sections_wrapper"> <div class="articleContent__ArticleWrapper-sc-2yl9jy-2 hAzjHP"> <div class="sc-chbbiW mBGSe article_header_class"> <p class="sc-kZmsYB bkiEfx article_type">研究文章|开放获取</p><div class="sc-RcBXQ ztvYg article_citation"> 体积 2011年 <span class="sc-fZwumE cUsFHx">|</年代p一个n>文章的ID 310762年 <span class="sc-fZwumE cUsFHx">|</年代p一个n><一个class="sc-htpNat bUhGXt link sc-fQejPQ duuwge" href="https://doi.org/10.1155/2011/310762" aria-label="Doi-link">https://doi.org/10.1155/2011/310762</一个></div> <div class="sc-LKuAh lcOiDF article_full_citation"> <div class="sc-hzNEM gsmOCv isHide"> <p>塞浦路斯Vrochidis, Ioannis Kompatsiaris, Ioannis佩特雷, ”<年代p一个nclass="adjust-article-svg-size">利用隐式的用户反馈来提高互动视频检索</年代p一个n>”,<i>多媒体的发展</我>, 卷。2011年, 文章的ID310762年, 18   页面, 2011年。 https://doi.org/10.1155/2011/310762</p></div> <button class="sc-hSdWYo etauus sc-iBEsjs kPshFw" color="">显示引用</button></div> <h1 class="sc-iRbamj jZOxOO sc-iSDuPN kEVvRd article_title adjust-article-svg-size heading-title">利用隐式的用户反馈来提高互动视频检索</h1></div> <div class="sc-jrIrqw eHMFfo article_authors"> <hr class="sc-cEvuZC XkrFe hr__author"> <span class=""><b>塞浦路斯Vrochidis</b><一个class="sc-htpNat bUhGXt link" href="mailto:" aria-label="Mail Option"><span class="sc-htoDjs kNoiZu sc-hjRWVT irNWJI icon-colored _2zt-1j0LmaF8UjTQAM413G icon-themeable" title="" font-size="16"></span></a><a class="sc-htpNat bUhGXt link" href="http://orcid.org/0000-0002-2505-9178" aria-label="Redirect to Doi" target="_blank"><img alt="" class="sc-EHOje iiusBg sc-iybRtq iQSObO" title="" role="presentation" src="data:image/svg+xml;base64,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" height="16"></a>,<年代up>1,2</年代up></年代p一个n><年代pan class="">Ioannis Kompatsiaris<一个class="sc-htpNat bUhGXt link" href="http://orcid.org/0000-0001-6447-9020" aria-label="Redirect to Doi" target="_blank"><img alt="" class="sc-EHOje iiusBg sc-iybRtq iQSObO" title="" role="presentation" src="data:image/svg+xml;base64,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" height="16"></a>,<年代up>1</年代up>和</年代p一个n><年代p一个nclass="">Ioannis佩特雷<年代up>2</年代up></年代p一个n><div class="sc-fHxwqH dMuGGA isHide"> <p><sup>1</年代up><年代p一个n>海勒斯研究和技术中心信息和远程信息处理研究所,6日荷航Charilaou-Thermi路,57001年的希腊塞萨洛尼基</年代p一个n></p><p><sup>2</年代up><年代p一个n>伦敦大学玛丽女王,英里路,英国伦敦E1 4 ns</年代p一个n></p></div> <button class="sc-hSdWYo etauus sc-iQtOjA kfDwZY" color="">显示更多</button></div> <div class="articleContent__EditorWrapper-sc-2yl9jy-3 evMkvU"> <div class="articleContent__TextWrapper-sc-2yl9jy-8 fYrLtk"> <b>学术编辑器:</b>安德里亚Prati</div> <div class="articleContent__PublicationTimeLineWrapper-sc-2yl9jy-4 fjviok"> <div class="articleContent__PublcationCol-sc-2yl9jy-5 hgKieC notPublished"> <span class="articleContent__PublicationField-sc-2yl9jy-6 cIribm">收到了</年代p一个n><年代p一个nclass="articleContent__PublicationValue-sc-2yl9jy-7 khHUcu">2010年9月01日</年代p一个n></div> <div class="articleContent__PublcationCol-sc-2yl9jy-5 hgKieC notPublished"> <span class="articleContent__PublicationField-sc-2yl9jy-6 cIribm">接受</年代p一个n><年代p一个nclass="articleContent__PublicationValue-sc-2yl9jy-7 khHUcu">2011年1月3日</年代p一个n></div> <div class="articleContent__PublcationCol-sc-2yl9jy-5 hgKieC isPublished"> <span class="articleContent__PublicationField-sc-2yl9jy-6 cIribm">发表</年代p一个n><年代p一个nclass="articleContent__PublicationValue-sc-2yl9jy-7 khHUcu">2011年3月24日</年代p一个n></div> </div> </div> <article class="sc-jtRlXQ gvEcCq article_body "> <div> <div class="xml-content"> <h4 class="header" id="abstract">文摘</h4><p>本文描述了一种方法来利用隐式的用户反馈收集在互动视频检索任务。我们提出一个框架,视频是第一个索引根据时间,文本和视觉特性,然后隐式使用基于用户反馈分析实现方法。生成的图像编码视频片段之间的语义关系基于过去的用户交互和随后用于生成的建议。此外,我们结合的视觉特性和隐式反馈信息通过训练支持向量机分类器的例子来自上述图以优化查询的视觉搜索示例。拟议的框架是评估进行真实用户实验。的结果表明,显著改善精度和召回报告后,利用隐式的用户反馈,同时提出了一种改进的排名在大多数视觉评估查询的例子。</p><年代p一个nclass="end-abs"></span> <h4 id="introduction">1。介绍</h4><p>近年来,数字技术的快速发展使得计算机的日益增长的存储和处理能力,以及快速的建立和先进的通信网络。考虑到同样的低成本图像和视频捕捉设备和互联网在当今社区的深层渗透,大量的视听内容已成为全球可用和可访问。这些内容的可用性和用户需要搜索到多媒体的需求集合的地方开发先进的多媒体搜索引擎;因此,视频检索的研究仍然是最具挑战性的任务之一。尽管最近在这个领域重大进展,进一步进步在几个视频检索领域需要提高当前的视频搜索引擎的性能。突破仍需要更具体地说,主要的研究领域的语义和互动搜索可能使用多通道分析和检索算法,以及相关反馈(<一个href="#B1">1</一个>]。</p><p>最先进的视频检索系统整合,结合一些先进技术包括文本检索和基于内容的视觉搜索,为了支持用户定位视频剪辑,满足他们的要求。这些方法所面临的主要挑战之一是生成有效的表征和描述的视频源。最初的一步这个方向是视频分割和索引分成更小的视频截图(<一个href="#B1">1</一个>]。基于视听信息,可以提取低层次特征(<一个href="#B2">2</一个>每一个镜头的);然而,由于众所周知的语义鸿沟问题,很难将它们与人类可以理解的概念。组合和融合的异构信息(如视觉、文本和运动)被认为是解决这个问题的第一步和有前景的结果呈现在图像和视频检索领域(<一个href="#B3">3</一个>- - - - - -<一个href="#B6">6</一个>]。然而,进展表示和索引的多媒体内容还没有设法克服语义鸿沟。</p><p>另一种方法是利用语义桥梁提供的隐式和显式反馈用户(<一个href="#B7">7</一个>的一个视频搜索引擎。在交互式视频检索任务,多个搜索会话,用户提交的查询、浏览视频信息,甚至提供明确的反馈结果。相关性反馈(RF)在信息检索机制被设计为互补的方法来进一步提高系统的性能要求显式反馈用户(即。识别正面和负面的例子),指导的机器学习技术。尽管图像和视频检索有前景的结果(<一个href="#B8">8</一个>- - - - - -<一个href="#B10">10</一个>),用户RF-based功能不是很受欢迎,因为他们的主要缺点是用户通常不愿意提供明确的信息(<一个href="#B11">11</一个>,<一个href="#B12">12</一个>]。这也反映在流行的web搜索引擎(如Google, YouTube等),提供用户界面,不请求或搜索会话期间需要显式的用户反馈。出于这一事实,我们建议利用隐式的用户反馈在搜索过程中,为了获得新的知识内容和涉及的多媒体项目之间建立语义关系。此信息可以利用生成的建议,以及改善现有的基于内容的检索方法的性能。</p><p>在这项工作中,我们认为隐式的用户反馈任何行动或导航行为的用户在互动视频检索任务,包括鼠标移动和点击和键盘输入和按键。使用隐式技术的主要优势是,他们不需要任何显式的用户反馈。虽然隐式信息一般被认为是比显式(不准确<一个href="#B13">13</一个>),大量的隐式数据(例如,在web搜索引擎日志文件)可以聚集在没有额外的努力给用户。最近在交互式视频搜索方法试图利用过去的用户交互执行检索(<一个href="#B7">7</一个>)或补充现有的基于内容的搜索模式(<一个href="#B14">14</一个>]。在这种背景下,我们提出一个视频检索框架,支持视频分析,以及隐含的用户反馈记录和处理。然后,我们根据过去的用户交互和提供建议我们提供混合视觉搜索模式相结合(即不同类地提取信息。隐含的反馈和视觉功能)采用机器学习的方法。</p><p>视频处理与应用程序实现文本、音频、图像和视频分析,包括视频镜头分割、关键帧提取和生成视觉功能,以及自动语音识别(ASR)和转换为文本。另一方面,我们利用隐式的用户反馈,为了启动语义视频片段之间的关系。这是由引入隐式视频搜索感兴趣的指标,然后通过构建语义关联图,它是利用生成的建议在以下两个步骤。第一,一个动作图描述生成用户导航模式采用一种新颖的方法,定义了搜索子(即。,部分会话的用户搜索一个特定的主题)基于查询分类。然后,一组行动图转化为一个加权图通过聚合操作图和分配权重量化的隐式的用户操作兴趣指标。为了提供建议,我们使用一个基于距离的算法来排名图节点。可视化查询的例子,这个图是利用以类似的方式来定义正面和负面的例子。后者与一组视觉合并相似和不同的例子基于视觉特性,为了构建一个训练集,用于训练支持向量机(SVM)分类器,reranks视觉搜索的结果。</p><p>这个框架实现了一个交互式视频搜索引擎,支持基本检索功能包括文本、视觉和时间搜索。搜索引擎使用的评价方法进行实际用户实验3个阶段:首先,一个基线系统仅支持视频分析检索选项使用用户,和他们的行为被记录;然后,不同的用户正在寻找话题比基线的使用稍微不同的增强版本的搜索引擎,也利用用户隐式反馈,为了评估建议;最后,在第三阶段中,不同的用户被雇来评估reranking的视觉效果。</p><p>详细论述了和整合的工作<一个href="#B15">15</一个>,<一个href="#B16">16</一个>到一个坚实的框架,而延长评估通过额外的实验和对比上述文章中介绍的技术。研究小说的贡献的工作中总结提出的基于查询的图表分析的方法分类和子的定义,以及方法结合视觉特性与隐式的用户反馈。作者所知,这是第一个尝试把过去的用户交互模式和视觉特性。</p><p>本文结构如下:部分<一个href="#sec2">2</一个>介绍了相关工作,在部分<一个href="#sec3">3</一个>,我们介绍了支持向量机用于reranking结果可视化查询的例子。视频索引和检索框架提出了部分<一个href="#sec4">4</一个>,而部分<一个href="#sec5">5</一个>处理视频内容分析。部分<一个href="#sec6">6</一个>描述用户隐式操作基于图的处理方法,和部分<一个href="#sec7">7</一个>提出的方法结合视觉特性与图结构的隐式的用户反馈。部分<一个href="#sec8">8</一个>展示了实现搜索引擎和用户交互模式,并给出了实验结果和评价在部分<一个href="#sec9">9</一个>。最后,部分<一个href="#sec10">10</一个>总结了纸。</p><h4我d="related-work">2。相关工作</h4><p>在一般情况下检索任务,隐式的用户反馈可以分为两大类:用户查询行为和物理反应。第一类包括用户与搜索引擎的交互的模式,作为一系列鼠标移动和点击,选择,关键的中风,和键盘输入,而第二个包括物理用户无意识行为眼球运动(例如,<一个href="#B17">17</一个>)、心率和大脑神经元的反应,可以聚集与脑电图(例如,<一个href="#B18">18</一个>])。一方面,第一类的反馈可以很容易地聚集即使在网络搜索会话,而生理反应可以借助特殊的记录可穿戴设备或其他传感器(如相机)捕获和分析用户行为。在这项工作中,我们将重点放在利用用户反馈,属于第一类。</p><p>隐式反馈方法基于用户与搜索引擎的交互的上下文中有效文本检索,他们大多是用于查询扩展和用户分析,以检索、过滤和推荐感兴趣的项目(<一个href="#B19">19</一个>]。介绍了“隐性利益指标”(<一个href="#B20">20.</一个>)作为一个特定的用户操作的定义可以被视为有意义的隐含的反馈。在[<一个href="#B21">21</一个>),作者进行了一个比较显式和隐式反馈系统得出结论,这两个系统之间没有显著差异,取代前者与后者可能是可行的。</p><p>在另一个有趣的工作<一个href="#B11">11</一个>),作者试图自动优化搜索引擎的检索质量使用点击数据。该方法利用点击数据(利用搜索引擎的查询日志)进行训练。支持向量机(SVM)方法,本文提出一种方法学习检索功能。更具体地说,点击数据翻译成排名用户首选项,然后用于火车检索功能。在这种情况下,支持向量机实现特别设计为了训练这样的排名,反映相关的用户首选项(即。,一种选择是比另一个)。</p><p>在[<一个href="#B12">12</一个>),提出检测(即“查询链”。,一个年代equence of queries) and then learn a retrieval function using SVMs. The authors demonstrate a simple method for automatically detecting query chains in query and clickthrough logs. These data are used to infer preference judgments regarding the relative relevance of documents both within individual query results and between documents returned by different queries within the same query chain. The method used to generate the preference judgments is validated using a controlled user study. A ranking SVM is adapted to learn a ranked retrieval function from the preference judgments. The results demonstrate significant improvements in the ranking given by a normal search engine.</p><p>隐式反馈技术尚未充分探讨了在多媒体领域(<一个href="#B22">22</一个>]。在文本检索,通常隐含的信息,可以考虑是用户选择(即。,theu年代erclicks on an interesting link or textual description to view the complete document), while in video retrieval, we have multiple interactions between the user and the system, which could be utilized to provide meaningful feedback. The main idea to exploit the user feedback during video retrieval interactive sessions is to extend the idea of “query chains” [<一个href="#B12">12</一个>),建立一个图表描述一系列用户操作。这样一个图转换为加权图通过聚合相同的节点之间的联系,和权重的基础上,介绍了不同的行为考虑在内。近期作品(<一个href="#B7">7</一个>,<一个href="#B23">23</一个>)使用上述技术来处理用户点击。</p><p>在[<一个href="#B7">7</一个>),提出使用基于社区的反馈开采前的用户交互的视频检索系统,基于欧卡皮鹿BM25检索模型支持文本查询来帮助用户在他们的搜索任务。这种反馈是为新用户提供建议的基础的视频检索系统。这是由代表所有用户交互与加权图。然后,这个隐式信息聚合来自多个会话和用户为一个代表,从而促进过去的隐式信息的分析和利用。在[<一个href="#B23">23</一个>),作者评估4种不同的算法,可以应用在这样加权图提供建议。然而,这些工作只考虑文本查询,而基本的视频检索选项视觉和基于时间的搜索将被忽略。此外,融合或结合隐式反馈数据与基于内容的方法不是未遂。</p><p>在另一个工作(<一个href="#B14">14</一个>),提出了一种视频检索系统,采用相关反馈和多通道融合不同来源(文本、视觉和点击数据),为了为用户产生推荐。在这种方法中,文本、视觉和听觉数据的视频镜头分别处理并与所选择的视频文件。然后,这些结果融合。进一步调整融合权重进行援助的点击数据,这表示用户的兴趣特定文档基于他/她的时间观看了视频拍摄。融合的方法和隐式反馈内容分析信息似乎是非常有趣的和有前途的;然而,在这个特定的工作,不是很深的隐式信息利用,查询操作的顺序是不考虑,在语义上互连后续查询和照片。该方法相比,这项工作利用隐式的用户反馈来调整权重,以融合结果从不同的搜索模式(例如,文本和视觉),而不是提供建议或改善的结果(即特定检索模块。在这种情况下,视觉搜索)。</p><h4我d="support-vector-machines">3所示。支持向量机</h4><p>支持向量机是一套监督学习方法,分析数据和识别模式和用来解决分类和回归问题。当一组训练正面和负面的例子,一个支持向量机训练算法构建一个模型,该模型预测的类别属于一个新的例子。实现这个目标,一个支持向量机结构超平面或一组超平面高或无限维空间。一般来说,人们认为最好的是通过分离超平面的最大距离最近的训练点的任何类。</p><p>在这项工作中,支持向量机用于reranking初始查询结果的视觉,视觉和隐式反馈信息的总和。我们使用一个支持向量机实现,实现了替代结构制定传统二进制分类的支持向量机优化问题描述的错误率(<一个href="#B24">24</一个>]。对于一个给定的训练集<年代vg height="14.6875" id="M1" style="vertical-align:-3.20526pt;width:122.225px;" version="1.1" viewbox="0 0 122.225 14.6875" width="122.225" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.6875)"> <g transform="translate(72,-60.25)"> <text transform="matrix(1,0,0,-1,-71.95,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="4.1634989" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-60.91,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> 1</t年代p一个n></text> <text transform="matrix(1,0,0,-1,-56.04,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="5.2137508" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-44.62,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> 1</t年代p一个n></text> <text transform="matrix(1,0,0,-1,-39.75,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> )</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="4.1634989" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="9.364747" y="0"> …</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="23.955748" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="29.156996" y="0"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="33.320496" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,0.46,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,5.31,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="5.2012482" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,16.72,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,21.57,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> )</t年代p一个n></text> </g> </g> </svg>与<年代vg height="17.549999" id="M2" style="vertical-align:-3.2316pt;width:54.612499px;" version="1.1" viewbox="0 0 54.612499 17.549999" width="54.612499" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,17.55)"> <g transform="translate(72,-57.96)"> <text transform="matrix(1,0,0,-1,-71.95,61.23)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-65.07,58.11)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-58.38,61.23)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ∈</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="12.027886" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-37.09,66.24)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>和<年代vg height="14.775" id="M3" style="vertical-align:-3.2316pt;width:89.849998px;" version="1.1" viewbox="0 0 89.849998 14.775" width="89.849998" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.775)"> <g transform="translate(72,-60.18)"> <text transform="matrix(1,0,0,-1,-71.95,63.45)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-65.75,60.32)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-59.06,63.45)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ∈</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="12.040389" y="0"> {</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="18.041828" y="0"> −</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="26.606384" y="0"> 1</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="32.857883" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="38.059132" y="0"> +</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="46.623688" y="0"> 1</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="52.875187" y="0"> }</t年代p一个n></text> </g> </g> </svg>、培训这个二进制分类支持向量机解决了以下优化问题,这对于预测提出了结构化输出和优化多变量性能的措施,如F<年代ub>1</年代ub>分数或精度/召回break-event点(<一个href="#B25">25</一个>]<年代p一个nclass="equation" id="EEq1"> <svg height="90.125" id="M4" preserveaspectratio="xMaxYMin" style="vertical-align:-0pt;width:700.9375px;" version="1.1" viewbox="0 0 700.9375 90.125" width="700.9375" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,90.125)"> <g transform="translate(72,0.1)"> <text transform="matrix(1,0,0,-1,150.64,56.25)"> <tspan style="font-size: 12.50px; " x="0" y="0"> 米</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="9.727334" y="0"> e</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="15.278666" y="0"> 一个</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="20.829998" y="0"> n</t年代p一个n></text> <text transform="matrix(1,0,0,-1,177.72,53.12)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="6.7828002" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="8.9708004" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="13.189264" y="0"> ≥</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="19.184383" y="0"> 0</t年代p一个n></text> <text transform="matrix(1,0,0,-1,202.98,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> 1</t年代p一个n></text> <g> <path d="m 202.98,59.37 6.25,0" style="fill:none;stroke:#000000;stroke-width:0.82499999;stroke-linecap:butt;stroke-linejoin:miter;stroke-miterlimit:10;stroke-opacity:1;stroke-dasharray:none"></path> </g> <text transform="matrix(1,0,0,-1,202.98,47.49)"> <tspan style="font-size: 12.50px; " x="0" y="0"> 2</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="7.4499998" y="-8.7600002"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,220.12,61.25)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,226.83,56.25)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="12.465491" y="0"> +</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="23.805712" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="32.195225" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="38.234173" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="247.28999" y="-0.34999999"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="251.45351" y="-0.34999999"> 1</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="257.70499" y="-0.34999999"> )</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="-157.09" y="40"> 年代</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="-152.22633" y="40"> u</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="-145.97484" y="40"> b</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="-139.72333" y="40"> j</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="-136.2475" y="40"> e</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="-130.69617" y="40"> c</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="-125.14484" y="40"> t</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="-117.5055" y="40"> t</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="-114.04217" y="40"> o</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="-104.32734" y="40"> ∶</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="-94.474976" y="40"> ∀</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="-87.473297" y="40"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="-78.808716" y="40"> ∈</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="-66.768326" y="40"> {</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="-60.766888" y="40"> 0</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="-54.515388" y="40"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="-49.301636" y="40"> 1</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="-43.050137" y="40"> }</t年代p一个n></text> <text transform="matrix(1,0,0,-1,189.78,21.48)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,198.1,16.25)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ∶</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="11.06" y="-7.25"> 1</t年代p一个n></text> <g> <path d="m 209.16,19.38 6.25,0" style="fill:none;stroke:#000000;stroke-width:0.82499999;stroke-linecap:butt;stroke-linejoin:miter;stroke-miterlimit:10;stroke-opacity:1;stroke-dasharray:none"></path> </g> <text transform="matrix(1,0,0,-1,209.18,7.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="7.4299998" y="-8.75"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,226.3,21.25)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="12.61" y="-10.94"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,233.01,9.7)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,234.54,2.52)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="2.7218721" y="0"> =</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="8.7169924" y="0"> 1</t年代p一个n></text> <text transform="matrix(1,0,0,-1,249.17,16.25)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,254.35,13.12)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,257.57,16.25)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,263.78,13.12)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,267,16.25)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,273.87,13.12)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,280.57,16.25)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ≥</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="13.23" y="-7.25"> 1</t年代p一个n></text> <g> <path d="m 293.8,19.38 6.25,0" style="fill:none;stroke:#000000;stroke-width:0.82499999;stroke-linecap:butt;stroke-linejoin:miter;stroke-miterlimit:10;stroke-opacity:1;stroke-dasharray:none"></path> </g> <text transform="matrix(1,0,0,-1,293.82,7.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,307.15,32.19)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,301.25,9.7)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,302.78,2.52)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="2.7218721" y="0"> =</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="8.7169924" y="0"> 1</t年代p一个n></text> <text transform="matrix(1,0,0,-1,317.4,16.25)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,322.59,13.12)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,328.59,16.25)"> <tspan style="font-size: 12.50px; " x="0" y="0"> −</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="11.340221" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="17.379169" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="145.53" y="0.14"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="149.6935" y="0.14"> 2</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="155.94499" y="0.14"> )</t年代p一个n></text> </g> </g> </svg></span>在哪里<年代vg height="10.6125" id="M6" style="vertical-align:-0.15048pt;width:11.8625px;" version="1.1" viewbox="0 0 11.8625 10.6125" width="11.8625" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,10.6125)"> <g transform="translate(72,-63.51)"> <text transform="matrix(1,0,0,-1,-71.95,63.71)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>常数和能力吗<年代vg height="7.4499998" id="M7" style="vertical-align:-0.11285pt;width:12.2375px;" version="1.1" viewbox="0 0 12.2375 7.4499998" width="12.2375" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,7.45)"> <g transform="translate(72,-66.04)"> <text transform="matrix(1,0,0,-1,-71.95,66.2)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>一个参数向量。这种方法有<年代vg height="11.4125" id="M8" style="vertical-align:-0.0pt;width:14px;" version="1.1" viewbox="0 0 14 11.4125" width="14" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,11.4125)"> <g transform="translate(72,-62.87)"> <text transform="matrix(1,0,0,-1,-71.95,62.92)"> <tspan style="font-size: 12.50px; " x="0" y="0"> 2</t年代p一个n></text> <text transform="matrix(1,0,0,-1,-65.7,68.09)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>限制,每个可能的一个向量<年代vg height="15.475" id="M9" style="vertical-align:-3.20526pt;width:155.3875px;" version="1.1" viewbox="0 0 155.3875 15.475" width="155.3875" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,15.475)"> <g transform="translate(72,-59.62)"> <text transform="matrix(1,0,0,-1,-71.95,62.87)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="8.6520758" y="0"> =</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="20.692465" y="0"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="24.855965" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-41.9,59.74)"> <tspan style="font-size: 8.75px; " x="0" y="0"> 1</t年代p一个n></text> <text transform="matrix(1,0,0,-1,-37.02,62.87)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="5.2012482" y="0"> …</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="19.79225" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="24.993498" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-6.83,59.74)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-1.98,62.87)"> <tspan style="font-size: 12.50px; " x="0" y="0"> )</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="7.6268301" y="0"> ∈</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="19.667219" y="0"> {</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="25.668659" y="0"> 0</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="31.920158" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="37.133911" y="0"> 1</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="43.38541" y="0"> }</t年代p一个n></text> <text transform="matrix(1,0,0,-1,47.41,68.09)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>,它只有一个松弛变量<年代vg height="13.425" id="M10" style="vertical-align:-2.29482pt;width:7.6750002px;" version="1.1" viewbox="0 0 7.6750002 13.425" width="7.6750002" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,13.425)"> <g transform="translate(72,-61.26)"> <text transform="matrix(1,0,0,-1,-71.95,63.6)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>共享的所有约束。的算法是用来解决上述分类支持向量机的优化问题是一个适应割平面算法。该算法迭代构造一个足够的子集<年代vg height="10.575" id="M11" style="vertical-align:-0.20064pt;width:17.9625px;" version="1.1" viewbox="0 0 17.9625 10.575" width="17.9625" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,10.575)"> <g transform="translate(72,-63.54)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>设置的限制。从一个空的一组约束<年代vg height="10.575" id="M12" style="vertical-align:-0.20064pt;width:23.1625px;" version="1.1" viewbox="0 0 23.1625 10.575" width="23.1625" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,10.575)"> <g transform="translate(72,-63.54)"> <text transform="matrix(1,0,0,-1,-67.78,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>在每个迭代中,首先计算最优在当前工作集<年代vg height="10.575" id="M13" style="vertical-align:-0.20064pt;width:17.9625px;" version="1.1" viewbox="0 0 17.9625 10.575" width="17.9625" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,10.575)"> <g transform="translate(72,-63.54)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>(例如,<年代vg height="10.9125" id="M14" style="vertical-align:-0.17555pt;width:39.4375px;" version="1.1" viewbox="0 0 39.4375 10.9125" width="39.4375" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,10.9125)"> <g transform="translate(72,-63.27)"> <text transform="matrix(1,0,0,-1,-71.95,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="13.153156" y="0"> =</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="25.193544" y="0"> 0</t年代p一个n></text> </g> </g> </svg>和<年代vg height="13.55" id="M15" style="vertical-align:-2.29482pt;width:34.875px;" version="1.1" viewbox="0 0 34.875 13.55" width="34.875" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,13.55)"> <g transform="translate(72,-61.16)"> <text transform="matrix(1,0,0,-1,-71.95,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="9.5022802" y="0"> =</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="21.542669" y="0"> 0</t年代p一个n></text> </g> </g> </svg>在第一次迭代),然后找到最违反约束的优化问题,并将其添加到工作集<年代vg height="10.575" id="M16" style="vertical-align:-0.20064pt;width:17.9625px;" version="1.1" viewbox="0 0 17.9625 10.575" width="17.9625" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,10.575)"> <g transform="translate(72,-63.54)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>(<一个href="#B24">24</一个>]。</p><p>选择这个特定的支持向量机实现的主要原因在这个方法是非常快的性能表明,这是重要的执行的排名优化实时查询过程中使用一个充分大的训练集。</p><h4我d="video-retrieval-framework">4所示。视频检索框架</h4><p>整个视频索引和检索框架提出了这种方法如图<一个href="//www.newsama.com/journals/am/2011/310762/fig1/" target="_blank">1</一个>。底部部分,视频进行分析,包括视频镜头分割,文本分析的音频和视觉信息提取的关键帧的处理。顶部框架的一部分,隐式的用户反馈是获取日志文件和分析为了启动部分的视听内容之间的语义关系。建议一代组件能够提供建议利用隐式的用户反馈处理结果。此外,隐含的反馈和视频分析的输出组合在一个单独的组件,采用支持向量机分类器的训练,在顺序重新排列结果可视化查询的例子。处理隐式信息和视频基于内容分析的脱机,而视觉的结合和隐式信息实现动态(即。实时提交一个新的查询时)。在接下来的部分,这些部分将详细描述。</p><div class="floats-partial-section partial-carousel-wrapper" data-section="fig1"> <div class="floats-partial-content fig-content-fig1"> <div class="partial-carousel--single-image"> <img alt="" name="310762.fig.001" src="https://static-01.hindawi.com/articles/am/volume-2011/310762/figures/310762.fig.001.jpg"> <br> <strong></strong> </div> </div> <div class="floats-partial-footer"> <div class="floats-partial-caption"> <span class="caption-text">图1</年代p一个n><一个一个r我一个-label="full view" class="caption-partial-url" href="//www.newsama.com/journals/am/2011/310762/fig1/" title="" target="_blank"></a> </div> <div class="floats-partial-caption-text"> <table> 视频索引和检索框架。</t一个ble> </div> </div> </div> <h4 id="video-analysis">5。视频分析</h4><p>视频分析部分的作用是处理最初的视频源,以生成和索引有效表征,可用于检索。时间、文本和visual-based索引操作执行所述的续集。</p><h5我d="sec5.1">5.1。时态索引</h5><p>正如简介中提到的,为了生成一个初始的高效表示视频源和索引它根据时间信息,镜头边界检测和镜头分割步骤所需的视频分割成镜头。在这个工作中,镜头检测是通过阈值颜色直方图对应两个连续帧之间的距离在一个视频<一个href="#B26">26</一个>]。然后,每个视频分割成更小的镜头基于镜头边界检测信息和中间的每一个镜头关键帧,这被认为是代表一个,提取。这样的话,时间索引结构构造,所以每一个视频都可以由一个图像序列(即表示。,一个图像/视频拍摄)。</p><h5我d="sec5.2">5.2。关键字索引</h5><p>索引的视频镜头后根据相关的文本信息是实现的方法<一个href="#B27">27</一个>]。音频信息处理与应用程序脱机自动语音识别(ASR)最初的视频源,这样特定的关键字可以分配给每一个镜头。索引和查询功能实现使用KinoSearch全文搜索引擎库(<一个href="#B28">28</一个>),创建一个二进制索引优化的速度。首先,stopwords从关键字删除文件之后,波特阻止算法的应用(<一个href="#B29">29日</一个>]。然后,任期为每个关键字权重计算通过应用BM25文本算法(<一个href="#B30">30.</一个>]。最后,关键字相关的分段。</p><h5我d="sec5.3">5.3。视觉相似索引</h5><p>视觉相似性镜头索引实现低水平的视觉描述符后的提取的方法<一个href="#B27">27</一个>]。在这种情况下,五MPEG-7描述符,即颜色布局、色彩结构、可伸缩的颜色、边缘直方图,均匀纹理,生成具有代表性的关键帧的每个视频拍摄和存储在关系数据库中。通过连接这些描述符,一个特征向量是制定简洁地表示多维空间中的每个关键帧。实证评估系统的性能使用不同的组合上述描述符提倡一个MPEG-7-based方案的选择,依靠颜色和质地(即。、颜色布局和边缘直方图连接)<一个href="#B27">27</一个>]。描述符之间的距离计算执行两枪采用MPEG实验模型中提出的功能(<一个href="#B31">31日</一个>]。</p><p>进一步提高系统的时间响应的性能,我们采用多维r - tree索引结构,构造脱机使用所有提取的关键帧的特征向量。r - tree (s) (<一个href="#B32">32</一个>)是适用于多维索引结构对象和已知促进大规模快速和高效检索。主成分分析(PCA)也用来降低初始空间的维数。</p><p>在查询阶段,由视觉的例子是发起查询时,查询的特征向量(即拍摄。,从它的具有代表性的关键帧)<年代vg height="12.925" id="M17" style="vertical-align:-1.90608pt;width:12.325px;" version="1.1" viewbox="0 0 12.325 12.925" width="12.325" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,12.925)"> <g transform="translate(72,-61.66)"> <text transform="matrix(1,0,0,-1,-71.95,63.61)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>是提取并提交r - tree索引结构。后者返回一组没有排名<年代vg height="10.325" id="M18" style="vertical-align:-0.0pt;width:13.2875px;" version="1.1" viewbox="0 0 13.2875 10.325" width="13.2875" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,10.325)"> <g transform="translate(72,-63.74)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>结果<年代vg height="14.4125" id="M19" style="vertical-align:-3.27548pt;width:21.0625px;" version="1.1" viewbox="0 0 21.0625 14.4125" width="21.0625" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.4125)"> <g transform="translate(72,-60.47)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-62.7,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>,其中包含<年代vg height="14.7125" id="M20" style="vertical-align:-3.2316pt;width:91.3125px;" version="1.1" viewbox="0 0 91.3125 14.7125" width="91.3125" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.7125)"> <g transform="translate(72,-60.23)"> <text transform="matrix(1,0,0,-1,-71.95,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-66.45,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-63.23,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="5.2012482" y="0"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="9.364747" y="0"> 0</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="19.092081" y="0"> ≤</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="31.132469" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="38.496738" y="0"> ≤</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="50.524624" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="60.05191" y="0"> )</t年代p一个n></text> </g> </g> </svg>发现类似的查询。最后的视觉排名是由计算视觉的距离<年代vg height="14.65" id="M21" style="vertical-align:-3.27548pt;width:21.4625px;" version="1.1" viewbox="0 0 21.4625 14.65" width="21.4625" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.65)"> <g transform="translate(72,-60.28)"> <text transform="matrix(1,0,0,-1,-71.95,63.6)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-65.3,60.47)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>之间的<年代vg height="12.925" id="M22" style="vertical-align:-1.90608pt;width:12.325px;" version="1.1" viewbox="0 0 12.325 12.925" width="12.325" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,12.925)"> <g transform="translate(72,-61.66)"> <text transform="matrix(1,0,0,-1,-71.95,63.61)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>和所有的<年代vg height="14.4125" id="M23" style="vertical-align:-3.27548pt;width:24.7125px;" version="1.1" viewbox="0 0 24.7125 14.4125" width="24.7125" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.4125)"> <g transform="translate(72,-60.47)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-62.7,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>,我们有<年代vg height="14.775" id="M24" style="vertical-align:-3.27548pt;width:129.2625px;" version="1.1" viewbox="0 0 129.2625 14.775" width="129.2625" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.775)"> <g transform="translate(72,-60.18)"> <text transform="matrix(1,0,0,-1,-71.95,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-65.3,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-57.8,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="4.1634989" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="13.928342" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="19.129589" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="23.018023" y="0"> )</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="30.657356" y="0"> =</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="42.697746" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-8.16,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-0.66,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="4.1634989" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="13.928342" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="19.129589" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,23.98,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,27.2,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> )</t年代p一个n></text> </g> </g> </svg>,在那里<年代vg height="14.65" id="M25" style="vertical-align:-3.27548pt;width:21.825001px;" version="1.1" viewbox="0 0 21.825001 14.65" width="21.825001" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.65)"> <g transform="translate(72,-60.28)"> <text transform="matrix(1,0,0,-1,-71.95,63.6)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-65.01,60.47)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>视觉距离的计算功能和吗<年代vg height="14.4125" id="M26" style="vertical-align:-3.27548pt;width:51.349998px;" version="1.1" viewbox="0 0 51.349998 14.4125" width="51.349998" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.4125)"> <g transform="translate(72,-60.47)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-66.45,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-59.76,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ∈</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="12.040389" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-38.47,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>。的排名查询<年代vg height="12.925" id="M27" style="vertical-align:-1.90608pt;width:12.325px;" version="1.1" viewbox="0 0 12.325 12.925" width="12.325" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,12.925)"> <g transform="translate(72,-61.66)"> <text transform="matrix(1,0,0,-1,-71.95,63.61)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>可以被描述为一个新的有序集吗<年代vg height="14.775" id="M28" style="vertical-align:-3.27548pt;width:140.0125px;" version="1.1" viewbox="0 0 140.0125 14.775" width="140.0125" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.775)"> <g transform="translate(72,-60.18)"> <text transform="matrix(1,0,0,-1,-71.95,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="9.2522202" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-52.17,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-41.2,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> =</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="12.027886" y="0"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="16.191385" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-19.49,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-14.6,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="5.2012482" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-3.89,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,0.72,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="5.2012482" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,11.43,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,16,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="5.2012482" y="0"> …</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="19.79225" y="0"> )</t年代p一个n></text> </g> </g> </svg>,在那里<年代vg height="14.775" id="M29" style="vertical-align:-3.27548pt;width:225.77499px;" version="1.1" viewbox="0 0 225.77499 14.775" width="225.77499" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.775)"> <g transform="translate(72,-60.18)"> <text transform="matrix(1,0,0,-1,-71.95,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-65.3,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-57.8,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="4.1634989" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="13.928342" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="19.129589" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="25.406096" y="0"> )</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="33.045429" y="0"> ≤</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="45.085819" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-6.06,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,1.44,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="4.1634989" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="13.928342" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="19.129589" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="25.006001" y="0"> )</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="32.645332" y="0"> ≤</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="44.685722" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,52.78,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,60.28,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="4.1634989" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="13.928342" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="19.129589" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="24.318335" y="0"> )</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="28.481834" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="33.695583" y="0"> …</t年代p一个n></text> </g> </g> </svg>的基数<年代vg height="14.4125" id="M30" style="vertical-align:-3.27548pt;width:37.875px;" version="1.1" viewbox="0 0 37.875 14.4125" width="37.875" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.4125)"> <g transform="translate(72,-60.47)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="9.2522202" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-52.17,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>元素<年代vg height="10.325" id="M31" style="vertical-align:-0.0pt;width:13.2875px;" version="1.1" viewbox="0 0 13.2875 10.325" width="13.2875" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,10.325)"> <g transform="translate(72,-63.74)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>。</p><h4我d="implicit-feedback-analysis">6。隐式反馈分析</h4><p>本节介绍了隐式的用户反馈的分析表达的鼠标点击和键盘输入。首先,我们介绍了隐式兴趣指标,我们将考虑视频搜索,然后我们构建行动图基于用户导航模式,最后我们总他们一个加权图,可以使用生成的视频镜头之间的距离。</p><h5我d="sec6.1">6.1。隐式兴趣指标</h5><p>我们的目标是定义隐式兴趣指标(<一个href="#B20">20.</一个>用户交互的测量方面,为了利用后者的信息携带的用户感知提出多媒体材料。</p><p>基于视频分析和索引提供的搜索功能模块,它超越经典的基于文本的查询和已经包含在现有系统(例如,<一个href="#B33">33</一个>]),我们将介绍以下最低的用户操作可以被视为主要的隐式视频检索兴趣指标。<年代p一个nclass="list"><span class="list-item"><span class="list-label">(1)</年代p一个n><年代p一个nclass="list-content">基于文本的查询(TQ):用户插入一个关键字和提交查询。在本例中,我们假设这个关键字满足查询的用户有一个非常高的概率。</年代p一个n></年代p一个n><span class="list-item"><span class="list-label">(2)</年代p一个n><年代p一个nclass="list-content">可视化查询(VQ):用户提交一个查询通过视觉的例子。我们假设当用户选择一个关键帧和视觉搜索类似的图片,然后也有兴趣使用的例子。</年代p一个n></年代p一个n><span class="list-item"><span class="list-label">(3)</年代p一个n><年代p一个nclass="list-content">侧覆查询(平方):用户选择一枪为了查看暂时相邻的照片和相关的文字描述。在这种情况下,用户很可能会感兴趣的选择。</年代p一个n></年代p一个n><span class="list-item"><span class="list-label">(4)</年代p一个n><年代p一个nclass="list-content">录像查询(VSQ):用户选择一枪(从另一个查询的结果),并检索所有的照片相同的视频。在这种情况下,我们认为用户感兴趣的初始一定的扩展。</年代p一个n></年代p一个n><span class="list-item"><span class="list-label">(5)</年代p一个n><年代p一个nclass="list-content">提交一个镜头(SS):用户标志着镜头有关。在本例中,我们假设用户这张照片很感兴趣。在网络搜索系统中,这个动作可以相当于看视频拍摄。</年代p一个n></年代p一个n></span></p> <h5 id="sec6.2">6.2。行动图</h5><p>我们利用隐式反馈信息通过使用一个扩展的变异图的建设方法提出了(<一个href="#B7">7</一个>]。虽然[<一个href="#B7">7</一个>]使用一个搜索系统仅支持基于文本的查询,我们处理一个更复杂的情况自提出交互式检索系统支持除了视觉和时间提交查询。</p><p>为了描述提出更好的方法,我们将介绍一些基本的定义。我们定义为“搜索会话”时期,某些用户花在搜索。我们认为“搜索子”时期某一用户花在寻找一个特定的主题。此外,我们提供了一个分类模式为用户操作在交互式视频搜索任务。首先,我们介绍“动词”的属性,描述一个动作根据其输出。更具体地说,我们认为一个行动是“传递”,当它生成一个输出和满足三联体<年代p一个nclass="equation" id="eq1"> <svg height="15.75" id="M32" preserveaspectratio="xMaxYMin" style="vertical-align:-0pt;width:700.9375px;" version="1.1" viewbox="0 0 700.9375 15.75" width="700.9375" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,15.75)"> <g transform="translate(72,-59.4)"> <text transform="matrix(1,0,0,-1,140.02,63.2)"> <tspan style="font-size: 12.50px; " x="0" y="0"> 我</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="3.4758339" y="0"> n</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="9.7148314" y="0"> p</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="15.966331" y="0"> u</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="22.217831" y="0"> t</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="31.244997" y="0"> →</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="48.374107" y="0"> 一个</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="53.925438" y="0"> c</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="59.476772" y="0"> t</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="62.952606" y="0"> 我</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="66.428436" y="0"> o</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="72.679939" y="0"> n</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="84.495277" y="0"> →</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="101.62438" y="0"> o</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="107.87589" y="0"> u</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="114.12738" y="0"> t</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="117.60322" y="0"> p</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="123.85472" y="0"> u</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="130.10622" y="0"> t</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="133.58205" y="0"> 。</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="334.10001" y="0"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="338.26349" y="0"> 3</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="344.51498" y="0"> )</t年代p一个n></text> </g> </g> </svg></span>另一方面,当一个行动不提供任何输出,(例如,<年代vg height="14.1875" id="M33" style="vertical-align:-2.72116pt;width:98.800003px;" version="1.1" viewbox="0 0 98.800003 14.1875" width="98.800003" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.1875)"> <g transform="translate(72,-60.65)"> <text transform="matrix(1,0,0,-1,-71.95,63.41)"> <tspan style="font-size: 12.50px; " x="0" y="0"> 我</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="3.4758339" y="0"> n</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="9.7148314" y="0"> p</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="15.966331" y="0"> u</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="22.217831" y="0"> t</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="31.244997" y="0"> →</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="48.374107" y="0"> 一个</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="53.925438" y="0"> c</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="59.476772" y="0"> t</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="62.952606" y="0"> 我</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="66.428436" y="0"> o</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="72.679939" y="0"> n</t年代p一个n></text> </g> </g> </svg>),它的特点是“不及物动词”。此外,我们分类查询操作分为两大类基于他们依赖与先前的行为:(a)自主查询,不依赖于以前的结果和(b)相关的查询,以作为输入结果从先前的搜索行动。</p><p>构建一个动作图根据用户搜索活动,我们利用涉及用户操作的属性。搜索会话期间,可以有一系列的传递行为,在一个动作的输出是输入的一部分,另一个(例如,从文本搜索结果视觉搜索的输入)。因此,创建一个动作图的两个节点之间的联系,我们需要有一个序列的两个动作,至少第一个传递。在搜索期间,用户可以搜索一个特定的主题;然而,它是可能的火灾搜索脑海中有非常广泛的或复杂的主题,甚至决定在会话期间更改搜索主题。出于这个原因,我们建议等会议不应该分析整体,但应该首先分解为子。假设每一个自主查询可以启动不同的主题搜索,我们提出一个新颖的方法,在此基础上,我们将每个搜索会话划分为“搜索子”那样生成几个子图和使用断点自主查询。</p><p>考虑相应的功能介绍隐性利益指标,只有文本搜索可以表示为自主查询,而另一个查询是相关的。在这种情况下,使用基于文本的查询之间的断点子见图的例子<一个href="//www.newsama.com/journals/am/2011/310762/fig2/" target="_blank">2</一个>。应该注意的是,在这个分析,可视化查询(VQ)并不构成一个自主查询,因为它是基于先前的查询的结果。然而,如果插入一个新的图像/视频支持,VQ可能也认为是自主。这些功能可以可视化的整体分类的两个不同的轴传递性和依赖性,如图<一个href="//www.newsama.com/journals/am/2011/310762/fig3/" target="_blank">3</一个>。</p><div class="floats-partial-section partial-carousel-wrapper" data-section="fig2"> <div class="floats-partial-content fig-content-fig2"> <div class="partial-carousel--single-image"> <img alt="" name="310762.fig.002" src="https://static-01.hindawi.com/articles/am/volume-2011/310762/figures/310762.fig.002.jpg"> <br> <strong></strong> </div> </div> <div class="floats-partial-footer"> <div class="floats-partial-caption"> <span class="caption-text">图2</年代p一个n><一个一个r我一个-label="full view" class="caption-partial-url" href="//www.newsama.com/journals/am/2011/310762/fig2/" title="" target="_blank"></a> </div> <div class="floats-partial-caption-text"> <table> 搜索会话和子。</t一个ble> </div> </div> </div> <div class="floats-partial-section partial-carousel-wrapper" data-section="fig3"> <div class="floats-partial-content fig-content-fig3"> <div class="partial-carousel--single-image"> <img alt="" name="310762.fig.003" src="https://static-01.hindawi.com/articles/am/volume-2011/310762/figures/310762.fig.003.jpg"> <br> <strong></strong> </div> </div> <div class="floats-partial-footer"> <div class="floats-partial-caption"> <span class="caption-text">图3</年代p一个n><一个一个r我一个-label="full view" class="caption-partial-url" href="//www.newsama.com/journals/am/2011/310762/fig3/" title="" target="_blank"></a> </div> <div class="floats-partial-caption-text"> <table> 用户操作的分类。</t一个ble> </div> </div> </div> <p>在一般情况下,一个搜索的子<年代vg height="10.725" id="M34" style="vertical-align:-0.1254pt;width:11.375px;" version="1.1" viewbox="0 0 11.375 10.725" width="11.375" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,10.725)"> <g transform="translate(72,-63.42)"> <text transform="matrix(1,0,0,-1,-71.95,63.6)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>由一组动作<年代vg height="14.575" id="M35" style="vertical-align:-3.2316pt;width:16.762501px;" version="1.1" viewbox="0 0 16.762501 14.575" width="16.762501" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.575)"> <g transform="translate(72,-60.34)"> <text transform="matrix(1,0,0,-1,-71.95,63.61)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-62.99,60.48)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>包括一个自治和一些相关的查询操作。该子图<年代vg height="14.5875" id="M36" style="vertical-align:-3.2316pt;width:17.0375px;" version="1.1" viewbox="0 0 17.0375 14.5875" width="17.0375" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.5875)"> <g transform="translate(72,-60.33)"> <text transform="matrix(1,0,0,-1,-71.95,63.6)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-62.77,60.47)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>由一组节点(即。,年代hot年代和keywords that represent inputs and outputs of a set of actions<年代vg height="14.575" id="M37" style="vertical-align:-3.2316pt;width:16.762501px;" version="1.1" viewbox="0 0 16.762501 14.575" width="16.762501" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.575)"> <g transform="translate(72,-60.34)"> <text transform="matrix(1,0,0,-1,-71.95,63.61)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-62.99,60.48)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>)和链接代表对应的动作<年代vg height="14.575" id="M38" style="vertical-align:-3.2316pt;width:48.025002px;" version="1.1" viewbox="0 0 48.025002 14.575" width="48.025002" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.575)"> <g transform="translate(72,-60.34)"> <text transform="matrix(1,0,0,-1,-71.95,63.61)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-65.67,60.48)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-58.98,63.61)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ∈</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="12.027886" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-37.98,60.48)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>,在那里<年代vg height="14.775" id="M39" style="vertical-align:-3.22282pt;width:105.4625px;" version="1.1" viewbox="0 0 105.4625 14.775" width="105.4625" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.775)"> <g transform="translate(72,-60.18)"> <text transform="matrix(1,0,0,-1,-71.95,63.45)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="7.3517642" y="0"> ∈</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="19.392153" y="0"> {</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="25.393593" y="0"> 1</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="31.645094" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="36.858845" y="0"> …</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="51.437344" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="56.651093" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-4.65,60.32)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,2.15,63.45)"> <tspan style="font-size: 12.50px; " x="0" y="0"> }</t年代p一个n></text> </g> </g> </svg>和<年代vg height="14.35" id="M40" style="vertical-align:-3.22282pt;width:21.924999px;" version="1.1" viewbox="0 0 21.924999 14.35" width="21.924999" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.35)"> <g transform="translate(72,-60.52)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-61.31,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>元素的基数吗<年代vg height="14.575" id="M41" style="vertical-align:-3.2316pt;width:16.762501px;" version="1.1" viewbox="0 0 16.762501 14.575" width="16.762501" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.575)"> <g transform="translate(72,-60.34)"> <text transform="matrix(1,0,0,-1,-71.95,63.61)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-62.99,60.48)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>。的行动图搜索会话由几个子图,反映了各自的子会话和父节点自主查询。</p><p>这些都是图的例子<一个href="//www.newsama.com/journals/am/2011/310762/fig4/" target="_blank">4</一个>,一个行动图搜索会话。这里,用户正在寻找照片,人们坐在桌子说进行描述。我们观察到的三个关键词是用来搜索(即开始。,t一个lk, sit, and dialogue) were considered as the parents for new subgraphs. Then, we construct a single-action graph aggregating the action graphs from the different user sessions. More specifically, all the nodes from the individual action graphs are mapped to single-action graph, and then all the action edges are mapped onto the same graph, generating in that way multiple links between the nodes. We observe that the three keywords that were used to start the search (i.e., talk, sit, and dialogue) are considered as the parents for new subgraphs, which correspond to different subsessions. In this way, concepts with different semantic meaning are not interconnected (e.g., “talk” with “sit”), while keywords with similar semantic meaning (i.e., “talk” and “dialogue”) are eventually interconnected due to the visual similarity between two shots in different subgraphs. Then, we construct a single-action graph aggregating the action graphs from the different user sessions.</p><div class="floats-partial-section partial-carousel-wrapper" data-section="fig4"> <div class="floats-partial-content fig-content-fig4"> <div class="partial-carousel--single-image"> <img alt="" name="310762.fig.004" src="https://static-01.hindawi.com/articles/am/volume-2011/310762/figures/310762.fig.004.jpg"> <br> <strong></strong> </div> </div> <div class="floats-partial-footer"> <div class="floats-partial-caption"> <span class="caption-text">图4</年代p一个n><一个一个r我一个-label="full view" class="caption-partial-url" href="//www.newsama.com/journals/am/2011/310762/fig4/" title="" target="_blank"></a> </div> <div class="floats-partial-caption-text"> <table> 行动图后用户交互。</t一个ble> </div> </div> </div> <h5 id="sec6.3">6.3。加权图</h5><p>建设一个行动图后,我们在以下三个步骤:生成加权图(a)相关结果与父查询,将在这方面不及物动作转换为传递,(b)之间的多个链接相同的节点坍塌成一个和(c)行动转化为权重。</p><p>最终的重量<年代vg height="7.4499998" id="M42" style="vertical-align:-0.11285pt;width:12.2375px;" version="1.1" viewbox="0 0 12.2375 7.4499998" width="12.2375" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,7.45)"> <g transform="translate(72,-66.04)"> <text transform="matrix(1,0,0,-1,-71.95,66.2)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>的链接<年代vg height="7.1374998" id="M43" style="vertical-align:-0.10033pt;width:7.8874998px;" version="1.1" viewbox="0 0 7.8874998 7.1374998" width="7.8874998" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,7.1375)"> <g transform="translate(72,-66.29)"> <text transform="matrix(1,0,0,-1,-71.95,66.44)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>两个节点之间<年代vg height="10.7375" id="M44" style="vertical-align:-0.13794pt;width:8.6000004px;" version="1.1" viewbox="0 0 8.6000004 10.7375" width="8.6000004" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,10.7375)"> <g transform="translate(72,-63.41)"> <text transform="matrix(1,0,0,-1,-71.95,63.6)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>和<年代vg height="7.1374998" id="M45" style="vertical-align:-0.10033pt;width:11.225px;" version="1.1" viewbox="0 0 11.225 7.1374998" width="11.225" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,7.1375)"> <g transform="translate(72,-66.29)"> <text transform="matrix(1,0,0,-1,-71.95,66.44)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>是由<年代p一个nclass="equation" id="EEq3"> <svg height="33.450001" id="M46" preserveaspectratio="xMaxYMin" style="vertical-align:-0pt;width:700.9375px;" version="1.1" viewbox="0 0 700.9375 33.450001" width="700.9375" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,33.45)"> <g transform="translate(72,-45.24)"> <text transform="matrix(1,0,0,-1,162.77,56.25)"> <tspan style="font-size: 12.50px; " x="72.019997" y="-7.25"> 1</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="11.765323" y="0"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="15.928822" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="22.142813" y="0"> )</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="29.782146" y="0"> =</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="41.810032" y="0"> 1</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="50.849701" y="0"> −</t年代p一个n></text> <g> <path d="m 226.16,59.37 23.5,0" style="fill:none;stroke:#000000;stroke-width:0.82499999;stroke-linecap:butt;stroke-linejoin:miter;stroke-miterlimit:10;stroke-opacity:1;stroke-dasharray:none"></path> </g> <text transform="matrix(1,0,0,-1,226.16,47.49)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="8.96" y="-0.0099999998"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="13.123499" y="-0.0099999998"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="19.33749" y="-0.0099999998"> )</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="24.700001" y="-8.7600002"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="247.96001" y="-8.7600002"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="252.1235" y="-8.7600002"> 4</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="258.375" y="-8.7600002"> )</t年代p一个n></text> </g> </g> </svg></span>在哪里<年代vg height="13.45" id="M47" style="vertical-align:-2.21957pt;width:26.9px;" version="1.1" viewbox="0 0 26.9 13.45" width="26.9" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,13.45)"> <g transform="translate(72,-61.24)"> <text transform="matrix(1,0,0,-1,-71.95,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="6.8766499" y="0"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="11.040149" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="17.254141" y="0"> )</t年代p一个n></text> </g> </g> </svg>是权重的总和的每个操作连接节点<年代vg height="10.7375" id="M48" style="vertical-align:-0.13794pt;width:8.6000004px;" version="1.1" viewbox="0 0 8.6000004 10.7375" width="8.6000004" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,10.7375)"> <g transform="translate(72,-63.41)"> <text transform="matrix(1,0,0,-1,-71.95,63.6)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>和<年代vg height="7.1374998" id="M49" style="vertical-align:-0.10033pt;width:11.225px;" version="1.1" viewbox="0 0 11.225 7.1374998" width="11.225" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,7.1375)"> <g transform="translate(72,-66.29)"> <text transform="matrix(1,0,0,-1,-71.95,66.44)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>。这个和表达<年代p一个nclass="equation" id="EEq4"> <svg height="38.1875" id="M50" preserveaspectratio="xMaxYMin" style="vertical-align:-0pt;width:700.9375px;" version="1.1" viewbox="0 0 700.9375 38.1875" width="700.9375" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,38.1875)"> <g transform="translate(72,-41.45)"> <text transform="matrix(1,0,0,-1,164.69,59.76)"> <tspan style="font-size: 12.50px; " x="40.84" y="6.5500002"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="8.9521484" y="0"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="13.115647" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="19.329638" y="0"> )</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="26.968971" y="0"> =</t年代p一个n></text> <text transform="matrix(1,0,0,-1,203.7,46.23)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="4.3935041" y="0"> ∈</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="10.388624" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,220.26,44.04)"> <tspan style="font-size: 6.25px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,223.51,59.76)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="8.7270937" y="0"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="12.890593" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="19.167099" y="0"> )</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="25.418598" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="250.61" y="0"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="254.7735" y="0"> 5</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="261.02499" y="0"> )</t年代p一个n></text> </g> </g> </svg></span>在哪里<年代vg height="9.9375" id="M51" style="vertical-align:-2.34499pt;width:8.4375px;" version="1.1" viewbox="0 0 8.4375 9.9375" width="8.4375" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,9.9375)"> <g transform="translate(72,-64.05)"> <text transform="matrix(1,0,0,-1,-71.95,66.44)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>是一个函数映射一个隐式的重量和每个行动<年代vg height="7.1750002" id="M52" style="vertical-align:-0.1254pt;width:7.9749999px;" version="1.1" viewbox="0 0 7.9749999 7.1750002" width="7.9749999" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,7.175)"> <g transform="translate(72,-66.26)"> <text transform="matrix(1,0,0,-1,-71.95,66.44)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>是一种行动,属于组的行为<年代vg height="14.575" id="M53" style="vertical-align:-3.2316pt;width:52.612499px;" version="1.1" viewbox="0 0 52.612499 14.575" width="52.612499" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.575)"> <g transform="translate(72,-60.34)"> <text transform="matrix(1,0,0,-1,-71.95,63.61)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-63.14,60.48)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-55.31,63.61)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ⊆</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="12.027886" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-34.31,60.48)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>组成不同的节点之间的联系<年代vg height="10.7375" id="M54" style="vertical-align:-0.13794pt;width:8.6000004px;" version="1.1" viewbox="0 0 8.6000004 10.7375" width="8.6000004" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,10.7375)"> <g transform="translate(72,-63.41)"> <text transform="matrix(1,0,0,-1,-71.95,63.6)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>和<年代vg height="7.1374998" id="M55" style="vertical-align:-0.10033pt;width:11.225px;" version="1.1" viewbox="0 0 11.225 7.1374998" width="11.225" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,7.1375)"> <g transform="translate(72,-66.29)"> <text transform="matrix(1,0,0,-1,-71.95,66.44)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>(<一个href="#B7">7</一个>]。</p><p>在分析部分<一个href="#sec6.1">6.1</一个>,我们分配指示值(0 - 10),量化引入隐性利益相关的兴趣水平指标(表<一个href="//www.newsama.com/journals/am/2011/310762/tab1/" target="_blank">1</一个>)。为了分配代表权重,我们要求10个用户测量水平对每个搜索行动的范围在0和10之间。结果(即。,the一个verage weights for each action) of this survey are presented in Table<一个href="//www.newsama.com/journals/am/2011/310762/tab1/" target="_blank">1</一个>。使用(<一个href="#EEq3">4</一个>)和(<一个href="#EEq4">5</一个>)和定义的值<年代vg height="9.9375" id="M56" style="vertical-align:-2.34499pt;width:8.4375px;" version="1.1" viewbox="0 0 8.4375 9.9375" width="8.4375" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,9.9375)"> <g transform="translate(72,-64.05)"> <text transform="matrix(1,0,0,-1,-71.95,66.44)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>,我们能够构建加权图。</p><div class="floats-partial-section partial-other-wrapper" data-section="tab1"> <div class="floats-partial-content fig-content-tab1"> <table class="table-group" id="tab1"> <tbody> <tr> <td> <table class="table"> <tbody> <tr> <td class="thead-hr" colspan="4"> <hr></td> </tr> <tr class="thead"> <td align="left">行动<年代vg height="14.7125" id="M57" style="vertical-align:-3.2316pt" version="1.1" viewbox="0 0 22.4 14.7125" width="22.4" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.7125)"> <g transform="translate(72,-60.23)"> <text transform="matrix(1,0,0,-1,-71.95,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="4.1634989" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-61.51,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-58.29,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> )</t年代p一个n></text> </g> </g> </svg></td> <td align="center"><span style="width: 30.0875px;"> <svg height="14.7125" id="M58" style="vertical-align:-3.2316pt" version="1.1" viewbox="0 0 30.0875 14.7125" width="30.0875" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.7125)"> <g transform="translate(72,-60.23)"> <text transform="matrix(1,0,0,-1,-71.95,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="6.1514759" y="0"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="10.314975" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-55.36,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-52.14,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> )</t年代p一个n></text> </g> </g> </svg></span></td> <td align="left">行动<年代vg height="14.7125" id="M59" style="vertical-align:-3.2316pt" version="1.1" viewbox="0 0 22.4 14.7125" width="22.4" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.7125)"> <g transform="translate(72,-60.23)"> <text transform="matrix(1,0,0,-1,-71.95,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="4.1634989" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-61.51,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-58.29,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> )</t年代p一个n></text> </g> </g> </svg></td> <td align="center"><span style="width: 30.0875px;"> <svg height="14.7125" id="M60" style="vertical-align:-3.2316pt" version="1.1" viewbox="0 0 30.0875 14.7125" width="30.0875" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.7125)"> <g transform="translate(72,-60.23)"> <text transform="matrix(1,0,0,-1,-71.95,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="6.1514759" y="0"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="10.314975" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-55.36,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-52.14,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> )</t年代p一个n></text> </g> </g> </svg></span></td> </tr> <tr> <td class="thead-hr" colspan="4"> <hr></td> </tr> <tr> <td align="left">基于文本的查询(TQ操作)</td> <td align="center">7.9</td> <td align="left">可视化查询(VQ)</td> <td align="center">8</td> </tr> <tr> <td align="left">侧覆查询(平方)</td> <td align="center">7.1</td> <td align="left">提交一枪(SS)</td> <td align="center">9。1</td> </tr> <tr> <td align="left">录像查询(VSQ)</td> <td align="center">5。8</td> <td align="left"></td> <td align="center"></td> </tr> <tr class="table-tr"> <td colspan="4"> <hr class="tbody-hr"></td> </tr> </tbody> </table></td> </tr> </tbody> </table> </div> <div class="floats-partial-footer"> <div class="floats-partial-caption"> <span class="caption-text">表1</年代p一个n><一个一个r我一个-label="full view" class="caption-partial-url" href="//www.newsama.com/journals/am/2011/310762/tab1/" title="" target="_blank"></a> </div> <div class="floats-partial-caption-text"> 平均分配重量为每个行动。</div> </div> </div> <p>图<一个href="//www.newsama.com/journals/am/2011/310762/fig5/" target="_blank">5</一个>说明了加权图,加工图的行动图后产生<一个href="//www.newsama.com/journals/am/2011/310762/fig4/" target="_blank">4</一个>根据上述方法。</p><div class="floats-partial-section partial-carousel-wrapper" data-section="fig5"> <div class="floats-partial-content fig-content-fig5"> <div class="partial-carousel--single-image"> <img alt="" name="310762.fig.005" src="https://static-01.hindawi.com/articles/am/volume-2011/310762/figures/310762.fig.005.jpg"> <br> <strong></strong> </div> </div> <div class="floats-partial-footer"> <div class="floats-partial-caption"> <span class="caption-text">图5</年代p一个n><一个一个r我一个-label="full view" class="caption-partial-url" href="//www.newsama.com/journals/am/2011/310762/fig5/" title="" target="_blank"></a> </div> <div class="floats-partial-caption-text"> <a href="//www.newsama.com/journals/am/2011/310762/fig4/">4</一个><t一个ble> 加权图后加工图的行动图。</t一个ble> </div> </div> </div> <h5 id="sec6.4">6.4。代的建议</h5><p>在[<一个href="#B23">23</一个>),几个推荐算法,提出了基于这样一个加权图。然而,在大多数的情况下,表现最好的算法是根据搜索主题。在这里,我们使用一个简单的算法,提升者建议基于加权图上的距离。后者计算两个节点之间的最短路径。两个不同的节点之间的距离的计算在这个图是用迪杰斯特拉算法的应用程序(<一个href="#B34">34</一个>),计算两个节点之间的路径短。尽管弗洛伊德的算法(<一个href="#B35">35</一个>)通常是更快的计算根据图论最短的距离,它更适合更密集的图形。在我们的示例中,生成的加权图被认为是相当稀少的,可以更有效地借助邻接列表代替邻接矩阵。因此,该方法被认为是可伸缩的,因为它可以处理巨大的和非常稀疏图,当许多用户和大型数据集。</p><p>随着计算距离是基于隐含的信息,但它揭示了语义关系,我们命名为“隐式语义距离”。因此,基于短路径的方法,我们可以计算每个查询的隐式语义距离<年代vg height="12.925" id="M61" style="vertical-align:-1.90608pt;width:12.325px;" version="1.1" viewbox="0 0 12.325 12.925" width="12.325" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,12.925)"> <g transform="translate(72,-61.66)"> <text transform="matrix(1,0,0,-1,-71.95,63.61)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>图中表示为一个节点,与其他节点包括在图中。在这种情况下,我们需要注意查询<年代vg height="12.925" id="M62" style="vertical-align:-1.90608pt;width:12.325px;" version="1.1" viewbox="0 0 12.325 12.925" width="12.325" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,12.925)"> <g transform="translate(72,-61.66)"> <text transform="matrix(1,0,0,-1,-71.95,63.61)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>可以一枪或一个关键字,而代表结果相同。在形式上,我们计算<年代vg height="14.7125" id="M63" style="vertical-align:-3.2316pt;width:124.2375px;" version="1.1" viewbox="0 0 124.2375 14.7125" width="124.2375" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.7125)"> <g transform="translate(72,-60.23)"> <text transform="matrix(1,0,0,-1,-71.95,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-65.3,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-59.81,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="4.1634989" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="13.928342" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="19.129589" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="23.018023" y="0"> )</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="30.657356" y="0"> =</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="42.697746" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-10.17,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-4.68,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="4.1634989" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="13.928342" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="19.129589" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,19.95,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,23.17,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> )</t年代p一个n></text> </g> </g> </svg>,在那里<年代vg height="14.475" id="M64" style="vertical-align:-3.13504pt;width:19.299999px;" version="1.1" viewbox="0 0 19.299999 14.475" width="19.299999" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.475)"> <g transform="translate(72,-60.42)"> <text transform="matrix(1,0,0,-1,-71.95,63.6)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-65.01,60.47)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>是隐式的距离计算功能,<年代vg height="14.3625" id="M65" style="vertical-align:-3.2316pt;width:48.837502px;" version="1.1" viewbox="0 0 48.837502 14.3625" width="48.837502" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.3625)"> <g transform="translate(72,-60.51)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-66.45,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-59.76,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ∈</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="12.040389" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-38.47,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>,<年代vg height="14.2375" id="M66" style="vertical-align:-3.13504pt;width:22.200001px;" version="1.1" viewbox="0 0 22.200001 14.2375" width="22.200001" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.2375)"> <g transform="translate(72,-60.61)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-62.7,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>是一组<年代vg height="10.325" id="M67" style="vertical-align:-0.0pt;width:17.237499px;" version="1.1" viewbox="0 0 17.237499 10.325" width="17.237499" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,10.325)"> <g transform="translate(72,-63.74)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>照片或/和关键词通过联系相互关联的查询<年代vg height="12.925" id="M68" style="vertical-align:-1.90608pt;width:12.325px;" version="1.1" viewbox="0 0 12.325 12.925" width="12.325" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,12.925)"> <g transform="translate(72,-61.66)"> <text transform="matrix(1,0,0,-1,-71.95,63.61)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>和<年代vg height="12.3" id="M69" style="vertical-align:-1.29163pt;width:68.6875px;" version="1.1" viewbox="0 0 68.6875 12.3" width="68.6875" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,12.3)"> <g transform="translate(72,-62.16)"> <text transform="matrix(1,0,0,-1,-71.95,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> 1</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="9.7148314" y="0"> ≤</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="21.75522" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="29.119488" y="0"> ≤</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="41.159878" y="0"> </t年代p一个n></text> </g> </g> </svg>。的排名查询的建议<年代vg height="12.925" id="M70" style="vertical-align:-1.90608pt;width:12.325px;" version="1.1" viewbox="0 0 12.325 12.925" width="12.325" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,12.925)"> <g transform="translate(72,-61.66)"> <text transform="matrix(1,0,0,-1,-71.95,63.61)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>可以被描述为一个新的有序集吗<年代vg height="14.9875" id="M71" style="vertical-align:-3.44714pt;width:137.5px;" version="1.1" viewbox="0 0 137.5 14.9875" width="137.5" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.9875)"> <g transform="translate(72,-60.01)"> <text transform="matrix(1,0,0,-1,-71.95,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="9.2522202" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-52.17,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-43.21,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> =</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="12.027886" y="0"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="16.191385" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-16.41,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> ́</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="-5.0936642" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-16.62,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="5.2012482" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-1.54,62.07)"> <tspan style="font-size: 8.75px; " x="0" y="0"> ́</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="-4.3600001" y="1.91"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-1.29,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="5.2012482" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,13.47,62.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> ̇</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="0.25" y="2"> ́</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="-4.0472322" y="2"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,13.99,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="5.2012482" y="0"> …</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="19.779745" y="0"> )</t年代p一个n></text> </g> </g> </svg>,在那里<年代vg height="18.0375" id="M72" style="vertical-align:-3.13504pt;width:219.0125px;" version="1.1" viewbox="0 0 219.0125 18.0375" width="219.0125" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,18.0375)"> <g transform="translate(72,-57.57)"> <text transform="matrix(1,0,0,-1,-71.95,60.75)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-65.3,57.62)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-59.81,60.75)"> <tspan style="font-size: 12.50px; " x="0" y="0"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="4.1634989" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="13.928342" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="26.418839" y="0"> ́</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="19.142094" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="25.418598" y="0"> )</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="33.045429" y="0"> ≤</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="45.085819" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-8.07,57.62)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-2.58,60.75)"> <tspan style="font-size: 12.50px; " x="25.370001" y="-2.72"> ́</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="0" y="0"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="4.1634989" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="13.928342" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="19.129999" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="25.006411" y="0"> )</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="32.645744" y="0"> ≤</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="44.67363" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,48.76,57.62)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,54.24,60.75)"> <tspan style="font-size: 12.50px; " x="0" y="0"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="4.1634989" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="13.928342" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="25.281067" y="0"> ́</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="19.142094" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="24.943485" y="0"> )</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="29.106983" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="34.320736" y="0"> …</t年代p一个n></text> </g> </g> </svg>的基数<年代vg height="10.325" id="M73" style="vertical-align:-0.0pt;width:17.237499px;" version="1.1" viewbox="0 0 17.237499 10.325" width="17.237499" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,10.325)"> <g transform="translate(72,-63.74)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>元素。</p><p>加权图的另一个重要的功能是,它可以用来显示新的搜索词建议通过计算输入关键词词的距离与其他关键词加权图。类似地,它还可以生成相关术语查询通过视觉的例子展示给用户的关键词发现近距离的查询加权图中枪。</p><h4我d="combining-visual-and-implicit---feedback-information">7所示。结合视觉和隐式反馈信息</h4><p>这部分的目的是重新排列查询的初步结果可视化例子利用加权图。虽然结果基于视觉描述符通常是相当满意的,在许多情况下,视觉搜索无法获取结果相同语义的混淆的类似的颜色或纹理语义无关的描述。作为讨论的部分<一个href="#sec5.3">5。3</一个>,视觉搜索中执行以下两个步骤:(i)通过提交查询描述符的r - tree结构和(2)排名返回计算结果可视化描述符之间的距离。这个想法是为了调整适当的排序函数的帮助下语义相关的照片为了强调更多的具体的视觉特性,可以为每个查询的重要性。预计训练一个分类器与语义正面和负面的例子从用户隐式反馈可以充分优化排序功能。更具体地说,我们训练一个分类器对于每个视觉示例查询采用作为训练集视觉相似和不同的例子,以及所产生的积极的和消极的样本隐含的反馈信息,以重新排列最初的视觉效果。在图<一个href="//www.newsama.com/journals/am/2011/310762/fig6/" target="_blank">6</一个>的整体算法给出了建议的方法。</p><div class="floats-partial-section partial-carousel-wrapper" data-section="fig6"> <div class="floats-partial-content fig-content-fig6"> <div class="partial-carousel--single-image"> <img alt="" name="310762.fig.006" src="https://static-01.hindawi.com/articles/am/volume-2011/310762/figures/310762.fig.006.jpg"> <br> <strong></strong> </div> </div> <div class="floats-partial-footer"> <div class="floats-partial-caption"> <span class="caption-text">图6</年代p一个n><一个一个r我一个-label="full view" class="caption-partial-url" href="//www.newsama.com/journals/am/2011/310762/fig6/" title="" target="_blank"></a> </div> <div class="floats-partial-caption-text"> <table> 算法结合视觉特性和隐式的用户反馈。</t一个ble> </div> </div> </div> <p>如图表所示,当一个查询的例子<年代vg height="12.925" id="M74" style="vertical-align:-1.90608pt;width:12.325px;" version="1.1" viewbox="0 0 12.325 12.925" width="12.325" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,12.925)"> <g transform="translate(72,-61.66)"> <text transform="matrix(1,0,0,-1,-71.95,63.61)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>提交,我们生产以下排名数据集的结果:两套吗<年代vg height="14.2375" id="M75" style="vertical-align:-3.13504pt;width:18.549999px;" version="1.1" viewbox="0 0 18.549999 14.2375" width="18.549999" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.2375)"> <g transform="translate(72,-60.61)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-62.7,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>和<年代vg height="14.2375" id="M76" style="vertical-align:-3.13504pt;width:18px;" version="1.1" viewbox="0 0 18 14.2375" width="18" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.2375)"> <g transform="translate(72,-60.61)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-63.14,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>(例如,related and unrelated shots, resp.) using the weighted graph that is constructed by processing the past user interaction and one set<年代vg height="14.4125" id="M77" style="vertical-align:-3.27548pt;width:21.0625px;" version="1.1" viewbox="0 0 21.0625 14.4125" width="21.0625" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.4125)"> <g transform="translate(72,-60.47)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-62.7,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>提供的r - tree结构,包括视觉镜头根据视觉特性相关的工作。随后,这些集合合并为了构造一个训练集<年代vg height="10.325" id="M78" style="vertical-align:-0.0pt;width:11.225px;" version="1.1" viewbox="0 0 11.225 10.325" width="11.225" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,10.325)"> <g transform="translate(72,-63.74)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>然后训练支持向量机分类器利用特征的视觉描述符。最后,我们采用<年代vg height="14.4125" id="M79" style="vertical-align:-3.27548pt;width:21.0625px;" version="1.1" viewbox="0 0 21.0625 14.4125" width="21.0625" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.4125)"> <g transform="translate(72,-60.47)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-62.7,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>(例如,the年代et的结果from visual search) as the test set, which is ranked according to the degrees of coefficients that are the output of the classifier and represent a similarity metric between each shot of the test set and the query. In Sections<一个href="#sec7.1">7.1</一个>和<一个href="#sec7.2">7.2</一个>,我们提供详细的训练集建设和支持向量机的训练。</p><h5我d="sec7.1">7.1。训练集构造</h5><p>就像描述的部分<一个href="#sec3">3</一个>为了培养一个二进制支持向量机分类器,我们需要确定一个合适的训练集<年代vg height="14.2375" id="M80" style="vertical-align:-3.13504pt;width:81.612503px;" version="1.1" viewbox="0 0 81.612503 14.2375" width="81.612503" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.2375)"> <g transform="translate(72,-60.61)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="10.339981" y="0"> =</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="22.380369" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-42.69,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-32.9,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ∪</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="10.527526" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-15.49,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>,在那里<年代vg height="14.2375" id="M81" style="vertical-align:-3.13504pt;width:17.487499px;" version="1.1" viewbox="0 0 17.487499 14.2375" width="17.487499" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.2375)"> <g transform="translate(72,-60.61)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-65.07,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>组积极的和吗<年代vg height="14.2375" id="M82" style="vertical-align:-3.13504pt;width:19.637501px;" version="1.1" viewbox="0 0 19.637501 14.2375" width="19.637501" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.2375)"> <g transform="translate(72,-60.61)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-65.07,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>负样本的集合。利用加权图,我们可以提取的阳性样本数量更接近查询开枪的负样本数量放置尽可能进一步在这个图。因此,我们创建积极的样本集<年代vg height="15.775" id="M83" style="vertical-align:-4.37273pt;width:73.175003px;" version="1.1" viewbox="0 0 73.175003 15.775" width="73.175003" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,15.775)"> <g transform="translate(72,-59.38)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-65.07,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="4.2009602" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="6.3889599" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-48.2,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> =</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="12.027886" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-26.91,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="4.2009602" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="6.3889599" y="0"> </t年代p一个n></text> </g> </g> </svg>,所有<年代vg height="16.137501" id="M84" style="vertical-align:-4.37273pt;width:180.95px;" version="1.1" viewbox="0 0 180.95 16.137501" width="180.95" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,16.1375)"> <g transform="translate(72,-59.09)"> <text transform="matrix(1,0,0,-1,-71.95,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-66.45,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-59.76,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ∈</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="12.040389" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-38.47,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="4.2009602" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="6.3889599" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-25.07,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="9.3772497" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-9.04,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-3.55,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="4.1634989" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="13.928342" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="19.129589" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="23.018023" y="0"> )</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="30.657356" y="0"> <</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="42.697746" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,45.8,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="4.2009602" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="6.3889599" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,55.15,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> t</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="2.4943199" y="0"> h</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="6.8703198" y="0"> r</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="9.7759838" y="0"> e</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="13.661872" y="0"> 年代</t年代p一个n></text> </g> </g> </svg>和<年代vg height="16.012501" id="M85" style="vertical-align:-4.37273pt;width:42.075001px;" version="1.1" viewbox="0 0 42.075001 16.012501" width="42.075001" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,16.0125)"> <g transform="translate(72,-59.19)"> <text transform="matrix(1,0,0,-1,-71.95,63.6)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-65.3,60.47)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="4.2009602" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="6.3889599" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-55.95,60.47)"> <tspan style="font-size: 8.75px; " x="0" y="0"> t</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="2.4943199" y="0"> h</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="6.8703198" y="0"> r</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="9.7759838" y="0"> e</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="13.661872" y="0"> 年代</t年代p一个n></text> </g> </g> </svg>是一个实验设置的阈值。同样,我们定义一组负面例子采用另一个距离阈值<年代vg height="16.012501" id="M86" style="vertical-align:-4.37273pt;width:44.700001px;" version="1.1" viewbox="0 0 44.700001 16.012501" width="44.700001" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,16.0125)"> <g transform="translate(72,-59.19)"> <text transform="matrix(1,0,0,-1,-71.95,63.6)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-65.3,60.47)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="4.2009602" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="6.3889599" y="0"> ℎ</t年代p一个n></text> <text transform="matrix(1,0,0,-1,-53.86,60.47)"> <tspan style="font-size: 8.75px; " x="0" y="0"> t</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="2.503072" y="0"> h</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="6.8790722" y="0"> r</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="9.7847357" y="0"> e</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="13.670624" y="0"> 年代</t年代p一个n></text> </g> </g> </svg>,我们考虑<年代vg height="15.775" id="M87" style="vertical-align:-4.37273pt;width:64.887497px;" version="1.1" viewbox="0 0 64.887497 15.775" width="64.887497" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,15.775)"> <g transform="translate(72,-59.38)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-65.07,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="4.2009602" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="6.3889599" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-46.48,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> =</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="12.027886" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-25.63,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>,所有<年代vg height="16.137501" id="M88" style="vertical-align:-4.37273pt;width:176.7375px;" version="1.1" viewbox="0 0 176.7375 16.137501" width="176.7375" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,16.1375)"> <g transform="translate(72,-59.09)"> <text transform="matrix(1,0,0,-1,-71.95,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-66.45,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-63.23,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ∈</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="12.027886" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="20.842501" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-31.85,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-26.36,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="5.2012482" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-14.5,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-9.01,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="4.1634989" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="13.928342" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="19.129589" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="23.018023" y="0"> )</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="30.657356" y="0"> ></t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="42.685242" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,40.34,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="4.2009602" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="6.3889599" y="0"> ℎ</t年代p一个n></text> <text transform="matrix(1,0,0,-1,51.78,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> t</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="2.4943199" y="0"> h</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="6.8703198" y="0"> r</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="9.7759838" y="0"> e</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="13.661872" y="0"> 年代</t年代p一个n></text> </g> </g> </svg>。在最好的情况下,这些照片不应该相互联系与查询图中(即拍摄。<年代vg height="14.6" id="M89" style="vertical-align:-3.13504pt;width:83.150002px;" version="1.1" viewbox="0 0 83.150002 14.6" width="83.150002" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.6)"> <g transform="translate(72,-60.32)"> <text transform="matrix(1,0,0,-1,-71.95,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-65.3,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-59.81,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="4.1634989" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="13.928342" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="19.129589" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="23.018023" y="0"> )</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="30.657356" y="0"> =</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="42.697746" y="0"> ∞</t年代p一个n></text> </g> </g> </svg>)。或者,我们可以选择一个预定义的数量的正面和负面的样本图像和应用所需的只有距离阈值。</p><p>最明显的方法可以简单地训练分类器<年代vg height="15.775" id="M90" style="vertical-align:-4.37273pt;width:25.475px;" version="1.1" viewbox="0 0 25.475 15.775" width="25.475" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,15.775)"> <g transform="translate(72,-59.38)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-65.07,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="4.2009602" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="6.3889599" y="0"> </t年代p一个n></text> </g> </g> </svg>和<年代vg height="15.775" id="M91" style="vertical-align:-4.37273pt;width:27.625px;" version="1.1" viewbox="0 0 27.625 15.775" width="27.625" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,15.775)"> <g transform="translate(72,-59.38)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-65.07,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="4.2009602" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="6.3889599" y="0"> </t年代p一个n></text> </g> </g> </svg>。然而,由于这一事实隐式反馈并不总是准确的,这种方法并不总是有效的。另一方面,视觉搜索通常能够检索相似关键帧,这表明一个视觉距离几乎为零。因此,为了减少这种影响,除了利用视觉的结果排名,质量好,我们在积极的样品的照片包括视觉上更接近查询示例采用一组实验阈值。此外,我们在负样本中包括的一些视觉效果非常远离查询图像视觉距离阈值再次考虑。</p><p>形式上,我们构建一套新的积极的样本<年代vg height="15.775" id="M92" style="vertical-align:-4.37273pt;width:115.575px;" version="1.1" viewbox="0 0 115.575 15.775" width="115.575" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,15.775)"> <g transform="translate(72,-59.38)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-65.07,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="5.0324001" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="7.2203999" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-47.37,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> =</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="12.027886" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-26.08,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="5.0324001" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="7.2203999" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-8.38,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ⊆</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="12.040389" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,12.91,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>,在那里<年代vg height="14.4125" id="M93" style="vertical-align:-3.27548pt;width:24.7125px;" version="1.1" viewbox="0 0 24.7125 14.4125" width="24.7125" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.4125)"> <g transform="translate(72,-60.47)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-62.7,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>是所有的视觉搜索结果,<年代vg height="16.137501" id="M94" style="vertical-align:-4.37273pt;width:176px;" version="1.1" viewbox="0 0 176 16.137501" width="176" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,16.1375)"> <g transform="translate(72,-59.09)"> <text transform="matrix(1,0,0,-1,-71.95,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-66.45,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-63.23,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ∈</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="12.027886" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-41.94,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="5.0324001" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="7.2203999" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-27.71,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="5.2012482" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="11.852844" y="0"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="16.016342" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="25.781185" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="30.994938" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="34.883369" y="0"> )</t年代p一个n></text> <text transform="matrix(1,0,0,-1,11.34,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,22.32,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> <</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="12.027886" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,41.01,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="5.0324001" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="7.2203999" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,51.18,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> t</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="2.503072" y="0"> h</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="6.8790722" y="0"> r</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="9.7847357" y="0"> e</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="13.670624" y="0"> 年代</t年代p一个n></text> </g> </g> </svg>和<年代vg height="16.012501" id="M95" style="vertical-align:-4.37273pt;width:43.112499px;" version="1.1" viewbox="0 0 43.112499 16.012501" width="43.112499" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,16.0125)"> <g transform="translate(72,-59.19)"> <text transform="matrix(1,0,0,-1,-71.95,63.6)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-65.3,60.47)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="5.0324001" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="7.2203999" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-55.12,60.47)"> <tspan style="font-size: 8.75px; " x="0" y="0"> t</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="2.503072" y="0"> h</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="6.8790722" y="0"> r</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="9.7847357" y="0"> e</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="13.670624" y="0"> 年代</t年代p一个n></text> </g> </g> </svg>是一个实验设置的阈值。随后,我们定义一组负样本<年代vg height="15.775" id="M96" style="vertical-align:-4.37273pt;width:119.8875px;" version="1.1" viewbox="0 0 119.8875 15.775" width="119.8875" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,15.775)"> <g transform="translate(72,-59.38)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-65.07,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="5.0324001" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="7.2203999" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-45.65,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> =</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="12.040389" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-24.36,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="5.0324001" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="7.2203999" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-4.93,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ⊆</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="12.027886" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,16.36,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>,所有<年代vg height="16.137501" id="M97" style="vertical-align:-4.37273pt;width:180.7625px;" version="1.1" viewbox="0 0 180.7625 16.137501" width="180.7625" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,16.1375)"> <g transform="translate(72,-59.09)"> <text transform="matrix(1,0,0,-1,-71.95,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-66.45,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-63.23,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ∈</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="12.027886" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-41.94,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="5.0324001" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="7.2203999" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-25.98,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="5.2012482" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-14.12,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-6.62,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="4.1634989" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="13.928342" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="19.129589" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="23.018023" y="0"> )</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="30.657356" y="0"> ></t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="42.685242" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,42.73,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="5.0324001" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="7.2203999" y="0"> ℎ</t年代p一个n></text> <text transform="matrix(1,0,0,-1,55,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> t</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="2.4943199" y="0"> h</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="6.8703198" y="0"> r</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="9.7759838" y="0"> e</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="13.661872" y="0"> 年代</t年代p一个n></text> </g> </g> </svg>。这些实验距离阈值可以设置为特定的值,总是<年代vg height="16.012501" id="M98" style="vertical-align:-4.37273pt;width:108.1125px;" version="1.1" viewbox="0 0 108.1125 16.012501" width="108.1125" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,16.0125)"> <g transform="translate(72,-59.19)"> <text transform="matrix(1,0,0,-1,-71.95,63.6)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-65.3,60.47)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="5.0324001" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="7.2203999" y="0"> ℎ</t年代p一个n></text> <text transform="matrix(1,0,0,-1,-53.03,60.47)"> <tspan style="font-size: 8.75px; " x="0" y="0"> t</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="2.503072" y="0"> h</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="6.8790722" y="0"> r</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="9.7847357" y="0"> e</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="13.670624" y="0"> 年代</t年代p一个n></text> <text transform="matrix(1,0,0,-1,-31.99,63.6)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ></t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="12.027886" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-13.3,60.47)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="5.0324001" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="7.2203999" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-3.12,60.47)"> <tspan style="font-size: 8.75px; " x="0" y="0"> t</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="2.4943199" y="0"> h</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="6.8703198" y="0"> r</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="9.7759838" y="0"> e</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="13.661872" y="0"> 年代</t年代p一个n></text> </g> </g> </svg>或者他们可以由用户手动调整手动辅助相结合的隐式信息和视觉数据根据用户需求。最后的训练集<年代p一个nclass="equation" id="EEq5"> <svg height="19.35" id="M99" preserveaspectratio="xMaxYMin" style="vertical-align:-0pt;width:700.9375px;" version="1.1" viewbox="0 0 700.9375 19.35" width="700.9375" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,19.35)"> <g transform="translate(72,-56.52)"> <text transform="matrix(1,0,0,-1,89.6,61.13)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="10.352484" y="0"> =</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="22.380369" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,118.87,58.01)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,128.66,61.13)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ∪</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="10.515023" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,146.06,58.01)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,158.27,61.13)"> <tspan style="font-size: 12.50px; " x="0" y="0"> =</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="12.04" y="2.51"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="17.889999" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,183.04,58.01)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="4.2009602" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="6.3889599" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,199.21,61.13)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ∪</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="10.527526" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,216.62,58.01)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="5.0324001" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="7.2203999" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,230.85,58.62)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="8.6300001" y="-2.51"> ∪</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="21.25" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="27.1" y="-2.51"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,264.82,58.01)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="4.2009602" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="6.3889599" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,282.72,61.13)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ∪</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="10.527526" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,300.13,58.01)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="5.0324001" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="7.2203999" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,316.09,58.62)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="7.9299998" y="-2.51"> 。</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="158.03" y="-2.51"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="162.1935" y="-2.51"> 6</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="168.44499" y="-2.51"> )</t年代p一个n></text> </g> </g> </svg></span></p> <h5 id="sec7.2">7.2。支持向量机分类器</h5><p>的培训和执行结果的reranking实时查询期间,我们必须选择一个快速实现可以提供合理的时间。执行培训的优势,在半自动版本的模块中,用户将能够优化组合过程中通过调整两个参与排名的权重(即。视觉和隐式),这将反映不同距离的定义阈值训练数据建设。当然除了实现,训练数据集的大小由能够保持低速度的一个重要因素。因为我们的目标是重新排列的结果,最初的想法是采用SVM(排名<一个href="#B11">11</一个>),这可能是由用户偏好之间两种不同的训练。然而,根据加权图是如何构建的,特别是定量分配行为的重量,很明显,接近示例照片的拍摄更被视为积极的样本而不是用户相对偏好声明。</p><p>因此,我们采用执行较快SVM分类器实现部分中描述<一个href="#sec3">3</一个>利用正负样本的特征的连接边缘直方图、颜色布局描述符。假设连接视觉描述符<年代vg height="14.8125" id="M100" style="vertical-align:-3.25793pt;width:125.95px;" version="1.1" viewbox="0 0 125.95 14.8125" width="125.95" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.8125)"> <g transform="translate(72,-60.15)"> <text transform="matrix(1,0,0,-1,-71.95,63.45)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="10.652556" y="0"> =</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="22.692945" y="0"> {</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="28.694386" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-36.92,60.32)"> <tspan style="font-size: 8.75px; " x="0" y="0"> 0</t年代p一个n></text> <text transform="matrix(1,0,0,-1,-32.05,63.45)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="5.2012482" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-20.51,60.32)"> <tspan style="font-size: 8.75px; " x="0" y="0"> 1</t年代p一个n></text> <text transform="matrix(1,0,0,-1,-15.64,63.45)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="5.2012482" y="0"> …</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="19.79225" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="25.006001" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,15.7,60.32)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,22.71,63.45)"> <tspan style="font-size: 12.50px; " x="0" y="0"> }</t年代p一个n></text> </g> </g> </svg>,然后(<一个href="#EEq1">2</一个>)转换成<年代p一个nclass="equation" id="EEq6"> <svg height="48.349998" id="M101" preserveaspectratio="xMaxYMin" style="vertical-align:-0pt;width:700.9375px;" version="1.1" viewbox="0 0 700.9375 48.349998" width="700.9375" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,48.35)"> <g transform="translate(72,-33.32)"> <text transform="matrix(1,0,0,-1,99.86,49.67)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ∀</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="7.0016799" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="15.653756" y="0"> ∈</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="27.694145" y="0"> {</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="33.695583" y="0"> 0</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="39.947086" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="45.148335" y="0"> 1</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="51.399834" y="0"> }</t年代p一个n></text> <text transform="matrix(1,0,0,-1,157.27,54.9)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,165.59,49.67)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ∶</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="11.06" y="-7.2600002"> 1</t年代p一个n></text> <g> <path d="m 176.65,52.8 6.25,0" style="fill:none;stroke:#000000;stroke-width:0.82499999;stroke-linecap:butt;stroke-linejoin:miter;stroke-miterlimit:10;stroke-opacity:1;stroke-dasharray:none"></path> </g> <text transform="matrix(1,0,0,-1,176.67,40.92)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="7.4299998" y="-8.75"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,193.79,54.67)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="12.61" y="-10.94"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,200.5,43.12)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,202.03,35.94)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="2.7218721" y="0"> =</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="8.7169924" y="0"> 1</t年代p一个n></text> <text transform="matrix(1,0,0,-1,216.65,49.67)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,221.84,46.55)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,225.06,49.67)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,231.26,46.55)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,234.48,49.67)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,241.67,46.55)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,248.37,49.67)"> <tspan style="font-size: 12.50px; " x="0" y="0"> ≥</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="13.23" y="-7.2600002"> 1</t年代p一个n></text> <g> <path d="m 261.6,52.8 6.25,0" style="fill:none;stroke:#000000;stroke-width:0.82499999;stroke-linecap:butt;stroke-linejoin:miter;stroke-miterlimit:10;stroke-opacity:1;stroke-dasharray:none"></path> </g> <text transform="matrix(1,0,0,-1,261.62,40.92)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,274.95,65.61)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,269.05,43.12)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,270.58,35.94)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="2.7218721" y="0"> =</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="8.7169924" y="0"> 1</t年代p一个n></text> <text transform="matrix(1,0,0,-1,285.2,49.67)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,290.39,46.55)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,296.39,49.67)"> <tspan style="font-size: 12.50px; " x="0" y="0"> −</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="11.340221" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="17.379169" y="0"> 。</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="177.73" y="0"> (</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="181.89349" y="0"> 7</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="188.145" y="0"> )</t年代p一个n></text> </g> </g> </svg></span>训练后的分类器<年代vg height="10.325" id="M102" style="vertical-align:-0.0pt;width:11.225px;" version="1.1" viewbox="0 0 11.225 10.325" width="11.225" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,10.325)"> <g transform="translate(72,-63.74)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>,我们提供测试集最初的结果<年代vg height="14.4125" id="M103" style="vertical-align:-3.27548pt;width:26.275px;" version="1.1" viewbox="0 0 26.275 14.4125" width="26.275" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.4125)"> <g transform="translate(72,-60.47)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-62.7,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>基于视觉描述符。这个测试集是最后排名基于之间的距离,计算每一个镜头和超平面,所构造的模型。</p><h4我d="lelantus-search-engine">8。LELANTUS搜索引擎</h4><p>在本节中,我们将介绍LELANTUS (LELANTUS在希腊神话中是一位泰坦的能力“移动而不被看见”这意味着引擎收集和处理隐式的用户反馈以透明的方式对用户)视频搜索引擎实现拟议的框架。首先,我们将描述的接口,然后我们将演示其功能通过用户交互模式。一个交互式演示LELANTUS视频搜索引擎可在:<一个href="http://mklab-services.iti.gr/lelantus" target="_blank">http://mklab-services.iti.gr/lelantus/</一个>。</p><h5我d="sec8.1">8.1。接口</h5><p>搜索引擎界面如图<一个href="//www.newsama.com/journals/am/2011/310762/fig7/" target="_blank">7</一个>。所有突出显示功能可以被记录在用户交互。一探究竟,我们观察到图形用户界面(GUI)是由两部分组成:左列,提供基于文本的搜索选项,和主要的容器,这项研究的结果发表在同一时间提供选项,通过视觉的例子和时间搜索查询。</p><div class="floats-partial-section partial-carousel-wrapper" data-section="fig7"> <div class="floats-partial-content fig-content-fig7"> <div class="partial-carousel--single-image"> <img alt="" name="310762.fig.007" src="https://static-01.hindawi.com/articles/am/volume-2011/310762/figures/310762.fig.007.jpg"> <br> <strong></strong> </div> </div> <div class="floats-partial-footer"> <div class="floats-partial-caption"> <span class="caption-text">图7</年代p一个n><一个一个r我一个-label="full view" class="caption-partial-url" href="//www.newsama.com/journals/am/2011/310762/fig7/" title="" target="_blank"></a> </div> <div class="floats-partial-caption-text"> <table> 搜索引擎接口。</t一个ble> </div> </div> </div> <p>在左栏允许用户输入一个关键字以火一个基于文本的搜索。提供两种不同的选择:(i)执行文本搜索利用人造硅视网膜信息和基于加权图(2)搜索和接受建议。下面这部分,提出了一些相关关键词为每个查询提交使用加权图。最后,左列包含一个篮子存储结构,用户可以存储结果他/她发现。</p><p>主容器的部分结果。六个不同的选项可用于每一个镜头:(i)来执行一个查询通过视觉的例子使用MPEG图像描述符点击代表,(2)将一枪标记为(即相关的话题。,年代ub米我t一个年代hot),(我ii) to view all the shots of the same video, (iv) to fire an image by example search using the relations of the user interaction weighted graph, and (v) to execute a hybrid search, which combines visual features and implicit user feedback and finally as discussed in Section<一个href="#sec7">7</一个>和(vi)查看暂时邻(即。,ne我ghboring) shots of a selected video shot with the associated textual transcription, as it is shown in Figure<一个href="//www.newsama.com/journals/am/2011/310762/fig8/" target="_blank">8</一个>。</p><div class="floats-partial-section partial-carousel-wrapper" data-section="fig8"> <div class="floats-partial-content fig-content-fig8"> <div class="partial-carousel--single-image"> <img alt="" name="310762.fig.008" src="https://static-01.hindawi.com/articles/am/volume-2011/310762/figures/310762.fig.008.jpg"> <br> <strong></strong> </div> </div> <div class="floats-partial-footer"> <div class="floats-partial-caption"> <span class="caption-text">图8</年代p一个n><一个一个r我一个-label="full view" class="caption-partial-url" href="//www.newsama.com/journals/am/2011/310762/fig8/" title="" target="_blank"></a> </div> <div class="floats-partial-caption-text"> <table> GUI显示颞邻近的照片和相关的ASR文本的镜头。</t一个ble> </div> </div> </div> <h5 id="sec8.2">8.2。交互模式</h5><p>在本节中,改善的结果,当隐含的反馈是考虑了不同的交互模式。在这些例子中,系统访问TRECVID 2008测试数据收集(详细介绍部分<一个href="#sec9">9</一个>)。隐式的用户反馈信息已经聚集在第一节中描述的实验阶段<一个href="#sec9.2">9。2</一个>。</p><p>首先,我们提出一个使用场景,用户搜索场景水体可见通过输入关键字“水”(图<一个href="//www.newsama.com/journals/am/2011/310762/fig9/" target="_blank">9</一个>)。作为文本检索提供的嘈杂的信息进行自动语音识别(ASR),只有一些结果描述水的场景。执行相同的查询利用图与过去(即交互数据。显然,建议),我们得到一个更好的结果(图<一个href="//www.newsama.com/journals/am/2011/310762/fig10/" target="_blank">10</一个>)。同时,系统输出项建议使用加权图(图<一个href="//www.newsama.com/journals/am/2011/310762/fig11/" target="_blank">11</一个>)。在这种情况下,查询关键字是“水”,大部分的推荐词似乎有很高的语义相似度与输入项。</p><div class="floats-partial-section partial-carousel-wrapper" data-section="fig9"> <div class="floats-partial-content fig-content-fig9"> <div class="partial-carousel--single-image"> <img alt="" name="310762.fig.009" src="https://static-01.hindawi.com/articles/am/volume-2011/310762/figures/310762.fig.009.jpg"> <br> <strong></strong> </div> </div> <div class="floats-partial-footer"> <div class="floats-partial-caption"> <span class="caption-text">图9</年代p一个n><一个一个r我一个-label="full view" class="caption-partial-url" href="//www.newsama.com/journals/am/2011/310762/fig9/" title="" target="_blank"></a> </div> <div class="floats-partial-caption-text"> <table> 用户提交的文本查询关键字“水”搜索的ASR成绩单。</t一个ble> </div> </div> </div> <div class="floats-partial-section partial-carousel-wrapper" data-section="fig10"> <div class="floats-partial-content fig-content-fig10"> <div class="partial-carousel--single-image"> <img alt="" name="310762.fig.0010" src="https://static-01.hindawi.com/articles/am/volume-2011/310762/figures/310762.fig.0010.jpg"> <br> <strong></strong> </div> </div> <div class="floats-partial-footer"> <div class="floats-partial-caption"> <span class="caption-text">图10</年代p一个n><一个一个r我一个-label="full view" class="caption-partial-url" href="//www.newsama.com/journals/am/2011/310762/fig10/" title="" target="_blank"></a> </div> <div class="floats-partial-caption-text"> <table> 结果从一个文本查询(关键字“水”)搜索加权图的援助。</t一个ble> </div> </div> </div> <div class="floats-partial-section partial-carousel-wrapper" data-section="fig11"> <div class="floats-partial-content fig-content-fig11"> <div class="partial-carousel--single-image"> <img alt="" name="310762.fig.0011" src="https://static-01.hindawi.com/articles/am/volume-2011/310762/figures/310762.fig.0011.jpg"> <br> <strong></strong> </div> </div> <div class="floats-partial-footer"> <div class="floats-partial-caption"> <span class="caption-text">图11</年代p一个n><一个一个r我一个-label="full view" class="caption-partial-url" href="//www.newsama.com/journals/am/2011/310762/fig11/" title="" target="_blank"></a> </div> <div class="floats-partial-caption-text"> <table> 一种基于文本的查询关键字建议。</t一个ble> </div> </div> </div> <p>在第二个使用场景,用户使用可视化的查询示例方法找到类似于一个给定的图像。随后,我们给出了设置当用户搜索结果的三种不同的形式:(a)基于内容的搜索使用的视觉特性,(b)基于建议利用过去的用户交互,以及(c)混合搜索相结合的视觉特性和隐式的用户反馈。系统的输出是数据可视化<一个href="//www.newsama.com/journals/am/2011/310762/fig12/" target="_blank">12</一个>,<一个href="//www.newsama.com/journals/am/2011/310762/fig13/" target="_blank">13</一个>,<一个href="//www.newsama.com/journals/am/2011/310762/fig14/" target="_blank">14</一个>分别对上述搜索口味,而这个特定的系统所产生的相关术语查询如图<一个href="//www.newsama.com/journals/am/2011/310762/fig15/" target="_blank">15</一个>。如果我们试图评论系统的可视化输出,似乎前12混合方法的结果是更好的质量比其他搜索模块的结果。更具体地说,91.67%(11/12)的精确报告混合形态,而视觉只有建立搜索达到精度为58.3% (7/12)。当然,混合方法的良好性能是由于精度高75%(9/12)(图报告的建议<一个href="//www.newsama.com/journals/am/2011/310762/fig13/" target="_blank">13</一个>),后者实际上是用于描述的训练集的建设部分<一个href="#sec7.1">7.1</一个>。</p><div class="floats-partial-section partial-carousel-wrapper" data-section="fig12"> <div class="floats-partial-content fig-content-fig12"> <div class="partial-carousel--single-image"> <img alt="" name="310762.fig.0012" src="https://static-01.hindawi.com/articles/am/volume-2011/310762/figures/310762.fig.0012.jpg"> <br> <strong></strong> </div> </div> <div class="floats-partial-footer"> <div class="floats-partial-caption"> <span class="caption-text">图12</年代p一个n><一个一个r我一个-label="full view" class="caption-partial-url" href="//www.newsama.com/journals/am/2011/310762/fig12/" title="" target="_blank"></a> </div> <div class="floats-partial-caption-text"> <table> 查询图像的例子。基于内容分析是用于这个查询。输入图片左边的角落。</t一个ble> </div> </div> </div> <div class="floats-partial-section partial-carousel-wrapper" data-section="fig13"> <div class="floats-partial-content fig-content-fig13"> <div class="partial-carousel--single-image"> <img alt="" name="310762.fig.0013" src="https://static-01.hindawi.com/articles/am/volume-2011/310762/figures/310762.fig.0013.jpg"> <br> <strong></strong> </div> </div> <div class="floats-partial-footer"> <div class="floats-partial-caption"> <span class="caption-text">图13</年代p一个n><一个一个r我一个-label="full view" class="caption-partial-url" href="//www.newsama.com/journals/am/2011/310762/fig13/" title="" target="_blank"></a> </div> <div class="floats-partial-caption-text"> <table> 查询图像的例子。关系的加权图是用来实现查询。输入图片左边的角落。</t一个ble> </div> </div> </div> <div class="floats-partial-section partial-carousel-wrapper" data-section="fig14"> <div class="floats-partial-content fig-content-fig14"> <div class="partial-carousel--single-image"> <img alt="" name="310762.fig.0014" src="https://static-01.hindawi.com/articles/am/volume-2011/310762/figures/310762.fig.0014.jpg"> <br> <strong></strong> </div> </div> <div class="floats-partial-footer"> <div class="floats-partial-caption"> <span class="caption-text">图14</年代p一个n><一个一个r我一个-label="full view" class="caption-partial-url" href="//www.newsama.com/journals/am/2011/310762/fig14/" title="" target="_blank"></a> </div> <div class="floats-partial-caption-text"> <table> 混合搜索相结合的视觉特性和隐式的用户反馈。输入图片左边的角落。</t一个ble> </div> </div> </div> <div class="floats-partial-section partial-carousel-wrapper" data-section="fig15"> <div class="floats-partial-content fig-content-fig15"> <div class="partial-carousel--single-image"> <img alt="" name="310762.fig.0015" src="https://static-01.hindawi.com/articles/am/volume-2011/310762/figures/310762.fig.0015.jpg"> <br> <strong></strong> </div> </div> <div class="floats-partial-footer"> <div class="floats-partial-caption"> <span class="caption-text">图15</年代p一个n><一个一个r我一个-label="full view" class="caption-partial-url" href="//www.newsama.com/journals/am/2011/310762/fig15/" title="" target="_blank"></a> </div> <div class="floats-partial-caption-text"> <table> 关键字建议查询图像的例子。一个在正确的输入图像。</t一个ble> </div> </div> </div> <p>在这个特定的例子中,似乎只有考虑到执行,当检索可视化描述符,低距离估计也对摄影也有类似的颜色,但没有语义相似的查询。另一方面,当混合排名,似乎隐式的用户反馈给了语义味道只排名的照片共享共同的颜色特征进行排名更低。建议这个词而言,大多数的建议的话(即。7 9)似乎是语义相关的所选图像(图<一个href="//www.newsama.com/journals/am/2011/310762/fig15/" target="_blank">15</一个>)。</p><h4我d="experiments-and-evaluation">9。实验和评价</h4><p>为了评估建议的方法,我们使用LELANTUS视频搜索引擎进行实验,对比不同的检索方式。评价实验分为三个阶段:训练阶段,隐式的用户反馈记录,生成的评估建议,最后的评估混合搜索方法。</p><h5我d="sec9.1">9.1。涉及的数据集和地面真理</h5><p>在这些实验中,我们用注释视频组TRECVID 2008。荷兰这个集合包括约100小时的视频(新闻杂志,科学新闻,新闻报道,纪录片,教育规划,和档案视频)。的视频被分割成大约30.000镜头(NIST)提供的镜头分割和提出了最具代表性的关键帧的搜索引擎。人造硅视网膜是特文特大学提供的<一个href="http://www.utwente.nl/" target="_blank">http://www.utwente.nl/</一个>);然而,由于错误,通常是在自动语音识别和机器翻译从荷兰到英国,他们不能被视为高度可靠。地面真理是由美国国家标准与技术研究院(NIST):<一个href="http://www.nist.gov/" target="_blank">http://www.nist.gov/</一个>),包括注释镜头24查询主题。这些查询主题的一部分被用于我们的实验。</p><h5我d="sec9.2">9.2。培训阶段</h5><p>在第一阶段(即。,thetr一个我n我ng phase), 24 search sessions, each lasting 15 minutes, took place, in which 24 users searched for the following 6 different topics (i.e., 4 different users for each topic) and their actions were recorded.<年代p一个nclass="list"><span class="list-item"><span class="list-label">(一)</年代p一个n><年代p一个nclass="list-content">找到的照片3或更少的人围坐在一张桌子边。</年代p一个n></年代p一个n><span class="list-item"><span class="list-label">(B)</年代p一个n><年代p一个nclass="list-content">发现的一个或更多的人在后台主要是树木和植物;没有道路或建筑是可见的。</年代p一个n></年代p一个n><span class="list-item"><span class="list-label">(C)</年代p一个n><年代p一个nclass="list-content">找到的一个或更多的人,可以看到身体的水。</年代p一个n></年代p一个n><span class="list-item"><span class="list-label">(D)</年代p一个n><年代p一个nclass="list-content">找到的一个女人跟相机在一次采访中位于indoors-no他人可见。</年代p一个n></年代p一个n><span class="list-item"><span class="list-label">(E)</年代p一个n><年代p一个nclass="list-content">找到一个或多个纸片的镜头,每一个与写作、打字、印刷,填充框架面积的一半以上。</年代p一个n></年代p一个n><span class="list-item"><span class="list-label">(F)</年代p一个n><年代p一个nclass="list-content">发现的一个人在街上,镜头说话。</年代p一个n></年代p一个n></span></p> <p>考虑到用户没有意识到的机制用于构建加权图,他们被要求搜索,因为他们通常做的(即。不考虑特定的策略),他们介绍了一些噪声通过执行随机的和不可预知的跳跃在搜索会话。</p><p>在这个阶段,(即基线版本。,reco米米endation generation and hybrid search were disabled) of LELANTUS was used. All the functionalities of Figure<一个href="//www.newsama.com/journals/am/2011/310762/fig7/" target="_blank">7</一个>所支持的基线系统记录在用户交互。然后,我们构造了加权图的基础上,提出方法。数值数据(例如,数量的节点,链接,等等)的加权图被发表在表<一个href="//www.newsama.com/journals/am/2011/310762/tab2/" target="_blank">2</一个>。</p><div class="floats-partial-section partial-other-wrapper" data-section="tab2"> <div class="floats-partial-content fig-content-tab2"> <table class="table-group" id="tab2"> <tbody> <tr> <td> <table class="table"> <tbody> <tr> <td class="thead-hr" colspan="5"> <hr></td> </tr> <tr class="thead"> <td align="left"></td> <td align="center">节点</td> <td align="center">照片</td> <td align="center">关键字</td> <td align="center">链接</td> </tr> <tr> <td class="thead-hr" colspan="5"> <hr></td> </tr> <tr> <td align="left">加权图</td> <td align="center">1298年</td> <td align="center">1229年</td> <td align="center">69年</td> <td align="center">2659年</td> </tr> <tr class="table-tr"> <td colspan="5"> <hr class="tbody-hr"></td> </tr> </tbody> </table></td> </tr> </tbody> </table> </div> <div class="floats-partial-footer"> <div class="floats-partial-caption"> <span class="caption-text">表2</年代p一个n><一个一个r我一个-label="full view" class="caption-partial-url" href="//www.newsama.com/journals/am/2011/310762/tab2/" title="" target="_blank"></a> </div> <div class="floats-partial-caption-text"> 数值统计加权图。</div> </div> </div> <h5 id="sec9.3">9.3。建议的评估</h5><p>在第二阶段,我们招募了4种不同的用户,寻找了以下4个主题:2个相关(但不相同)的第一部分和2无关紧要。在本例中,每个用户搜索所有的4个主题。<年代p一个nclass="list"><span class="list-item"><span class="list-label">(1)</年代p一个n><年代p一个nclass="list-content">发现的一个或更多的人与一个或多个马。</年代p一个n></年代p一个n><span class="list-item"><span class="list-label">(2)</年代p一个n><年代p一个nclass="list-content">找到的照片(即地图。,类似于主题(E))。</年代p一个n></年代p一个n><span class="list-item"><span class="list-label">(3)</年代p一个n><年代p一个nclass="list-content">发现的一个或更多的人与一个或更多的书。</年代p一个n></年代p一个n><span class="list-item"><span class="list-label">(4)</年代p一个n><年代p一个nclass="list-content">找到照片的食物或饮料(即表。,类似于主题(A))。</年代p一个n></年代p一个n></span></p> <p>在这些搜索会话,用户使用LELANTUS的基线版本。然后,另一个4用户执行一个搜索同样的主题(1)- (4)。这些用户能够使用不仅系统的基本检索选项,但也(即推荐功能。增强版)。每个会话的时间是10分钟过去两个实验阶段。</p><p>性能方面的精度和召回的第二和第三阶段试验报告在表<一个href="//www.newsama.com/journals/am/2011/310762/tab3/" target="_blank">3</一个>和数据所示<一个href="//www.newsama.com/journals/am/2011/310762/fig16/" target="_blank">16</一个>和<一个href="//www.newsama.com/journals/am/2011/310762/fig17/" target="_blank">17</一个>分别计算这些指标对注释的结果为每个主题。</p><div class="floats-partial-section partial-other-wrapper" data-section="tab3"> <div class="floats-partial-content fig-content-tab3"> <table class="table-group" id="tab3"> <tbody> <tr> <td> <table class="table"> <tbody> <tr> <td class="thead-hr" colspan="5"> <hr></td> </tr> <tr> <td align="left" rowspan="2">主题</td> <td align="center" colspan="2">精度</td> <td align="center" colspan="2">回忆</td> </tr> <tr> <td align="center">基线</td> <td align="center">增强</td> <td align="center">基线</td> <td align="center">增强</td> </tr> <tr> <td align="center" colspan="5"> <hr></td> </tr> <tr> <td align="left">1</td> <td align="center">0.955556</td> <td align="center">0.896296</td> <td align="center">0.042042</td> <td align="center">0.045045</td> </tr> <tr> <td align="left">2</td> <td align="center">0.885467</td> <td align="center">0.900362</td> <td align="center">0.130719</td> <td align="center">0.134804</td> </tr> <tr> <td align="left">3</td> <td align="center">0.597154</td> <td align="center">0.664683</td> <td align="center">0.07906</td> <td align="center">0.096154</td> </tr> <tr> <td align="left">4</td> <td align="center">0.707702</td> <td align="center">0.766934</td> <td align="center">0.033712</td> <td align="center">0.075335</td> </tr> <tr class="table-tr"> <td colspan="5"> <hr class="tbody-hr"></td> </tr> </tbody> </table></td> </tr> </tbody> </table> </div> <div class="floats-partial-footer"> <div class="floats-partial-caption"> <span class="caption-text">表3</年代p一个n><一个一个r我一个-label="full view" class="caption-partial-url" href="//www.newsama.com/journals/am/2011/310762/tab3/" title="" target="_blank"></a> </div> <div class="floats-partial-caption-text"> 基线和增强系统的精度和召回。</div> </div> </div> <div class="floats-partial-section partial-carousel-wrapper" data-section="fig16"> <div class="floats-partial-content fig-content-fig16"> <div class="partial-carousel--single-image"> <img alt="" name="310762.fig.0016" src="https://static-01.hindawi.com/articles/am/volume-2011/310762/figures/310762.fig.0016.jpg"> <br> <strong></strong> </div> </div> <div class="floats-partial-footer"> <div class="floats-partial-caption"> <span class="caption-text">图16</年代p一个n><一个一个r我一个-label="full view" class="caption-partial-url" href="//www.newsama.com/journals/am/2011/310762/fig16/" title="" target="_blank"></a> </div> <div class="floats-partial-caption-text"> <table> 精密基线和增强系统的结果。</t一个ble> </div> </div> </div> <div class="floats-partial-section partial-carousel-wrapper" data-section="fig17"> <div class="floats-partial-content fig-content-fig17"> <div class="partial-carousel--single-image"> <img alt="" name="310762.fig.0017" src="https://static-01.hindawi.com/articles/am/volume-2011/310762/figures/310762.fig.0017.jpg"> <br> <strong></strong> </div> </div> <div class="floats-partial-footer"> <div class="floats-partial-caption"> <span class="caption-text">图17</年代p一个n><一个一个r我一个-label="full view" class="caption-partial-url" href="//www.newsama.com/journals/am/2011/310762/fig17/" title="" target="_blank"></a> </div> <div class="floats-partial-caption-text"> <table> 回忆的基线和增强系统的结果。</t一个ble> </div> </div> </div> <p>的平均改善回忆前两个话题(1)和(2)(即。,the我rrelevant to the initial ones) is about 5%, while precision seems to slightly drop by an average of 2%. These results can be considered rather reasonable due to the fact that past users have not interacted much with the video shots that are related to these topics. As expected, the major improvement is reported in the topics (3) and (4) (i.e., the relevant to the initial queries), in which recall and precision are increased by an average of 72% and 9.8%, respectively. The low absolute recall values are due to the fact that the shots that were annotated as relevant for each query-topic could not possibly be retrieved in the requested time duration of the experimental search sessions of this phase.</p><h5我d="sec9.4">9.4。视觉搜索优化实验</h5><p>在这个实验阶段,我们试图评估混合搜索模式,结合了视觉特性和隐式的用户反馈。</p><p>测量和评估提出算法的性能,我们比较两种不同的排名查询通过视觉的例子:(a)视觉排名,这是距离的视觉提供的描述符和(b)混合排名,这是后生成该算法的应用。比较上述检索功能,我们利用方法建议<一个href="#B11">11</一个>]。考虑到<年代vg height="14.2375" id="M104" style="vertical-align:-3.13504pt;width:39.650002px;" version="1.1" viewbox="0 0 39.650002 14.2375" width="39.650002" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.2375)"> <g transform="translate(72,-60.61)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="9.2522202" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-52.17,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>混合排名和吗<年代vg height="14.4125" id="M105" style="vertical-align:-3.27548pt;width:37.875px;" version="1.1" viewbox="0 0 37.875 14.4125" width="37.875" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.4125)"> <g transform="translate(72,-60.47)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="9.2522202" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-52.17,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>视觉,我们构造一个“综合排名”<年代vg height="15.775" id="M106" style="vertical-align:-4.37273pt;width:46.512501px;" version="1.1" viewbox="0 0 46.512501 15.775" width="46.512501" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,15.775)"> <g transform="translate(72,-59.38)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="9.2522202" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-52.17,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="7.6404958" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="9.828496" y="0"> </t年代p一个n></text> </g> </g> </svg>包括顶部链接的排名。更具体地说,<年代vg height="15.775" id="M107" style="vertical-align:-4.37273pt;width:46.512501px;" version="1.1" viewbox="0 0 46.512501 15.775" width="46.512501" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,15.775)"> <g transform="translate(72,-59.38)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="9.2522202" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-52.17,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="7.6404958" y="0"> ,</t年代p一个n><t年代p一个n style="font-size: 8.75px; " x="9.828496" y="0"> </t年代p一个n></text> </g> </g> </svg>由<年代vg height="10.725" id="M108" style="vertical-align:-0.1254pt;width:5.4124999px;" version="1.1" viewbox="0 0 5.4124999 10.725" width="5.4124999" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,10.725)"> <g transform="translate(72,-63.42)"> <text transform="matrix(1,0,0,-1,-71.95,63.6)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>结果,这实际上是<年代vg height="14.5875" id="M109" style="vertical-align:-3.22282pt;width:14.7125px;" version="1.1" viewbox="0 0 14.7125 14.5875" width="14.7125" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.5875)"> <g transform="translate(72,-60.33)"> <text transform="matrix(1,0,0,-1,-71.95,63.6)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-65.17,60.47)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>上面的结果<年代vg height="14.2375" id="M110" style="vertical-align:-3.13504pt;width:36.012501px;" version="1.1" viewbox="0 0 36.012501 14.2375" width="36.012501" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.2375)"> <g transform="translate(72,-60.61)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="9.2522202" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-52.17,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>和<年代vg height="14.5875" id="M111" style="vertical-align:-3.2316pt;width:14.3625px;" version="1.1" viewbox="0 0 14.3625 14.5875" width="14.3625" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.5875)"> <g transform="translate(72,-60.33)"> <text transform="matrix(1,0,0,-1,-71.95,63.6)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-65.17,60.47)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>的<年代vg height="14.4125" id="M112" style="vertical-align:-3.27548pt;width:34.224998px;" version="1.1" viewbox="0 0 34.224998 14.4125" width="34.224998" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.4125)"> <g transform="translate(72,-60.47)"> <text transform="matrix(1,0,0,-1,-71.95,63.79)"> <tspan style="font-size: 12.50px; " x="0" y="0"> </t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="9.2522202" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-52.17,60.66)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> </g> </g> </svg>,在那里<年代vg height="14.7125" id="M113" style="vertical-align:-3.2316pt;width:80.050003px;" version="1.1" viewbox="0 0 80.050003 14.7125" width="80.050003" xmlns="http://www.w3.org/2000/svg"> <g transform="matrix(1.25,0,0,-1.25,0,14.7125)"> <g transform="translate(72,-60.23)"> <text transform="matrix(1,0,0,-1,-71.95,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> |</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="2.5006001" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-62.67,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-55,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> −</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="11.340221" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-36.88,60.37)"> <tspan style="font-size: 8.75px; " x="0" y="0"> </t年代p一个n></text> <text transform="matrix(1,0,0,-1,-32.27,63.5)"> <tspan style="font-size: 12.50px; " x="0" y="0"> |</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="5.9639311" y="0"> ≤</t年代p一个n><t年代p一个n style="font-size: 12.50px; " x="18.00432" y="0"> 1</t年代p一个n></text> </g> </g> </svg>。这种形式的演讲导致盲目的统计检验,以便用户选择,表示为点击排名的结果,证明了公正的偏好。这些方法可以被认为是更适当的应用于视觉实例查询而不是文本搜索,在这种情况下,结果不是很主观的用户。数据<一个href="//www.newsama.com/journals/am/2011/310762/fig12/" target="_blank">12</一个>和<一个href="//www.newsama.com/journals/am/2011/310762/fig14/" target="_blank">14</一个>分别说明视觉和混合排名,而在图<一个href="//www.newsama.com/journals/am/2011/310762/fig18/" target="_blank">18</一个>相同的综合排名查询被描述。</p><div class="floats-partial-section partial-carousel-wrapper" data-section="fig18"> <div class="floats-partial-content fig-content-fig18"> <div class="partial-carousel--single-image"> <img alt="" name="310762.fig.0018" src="https://static-01.hindawi.com/articles/am/volume-2011/310762/figures/310762.fig.0018.jpg"> <br> <strong></strong> </div> </div> <div class="floats-partial-footer"> <div class="floats-partial-caption"> <span class="caption-text">图18</年代p一个n><一个一个r我一个-label="full view" class="caption-partial-url" href="//www.newsama.com/journals/am/2011/310762/fig18/" title="" target="_blank"></a> </div> <div class="floats-partial-caption-text"> <table> 综合排名结果的查询在左上角。</t一个ble> </div> </div> </div> <p>在此阶段,12个用户用来确认视觉相似的结果为200查询视觉加权图中包含的例子。构建有效的训练集,选择阈值与实验,而30积极和30的负面例子被每个查询的加权图中提取。在表中<一个href="//www.newsama.com/journals/am/2011/310762/tab4/" target="_blank">4</一个>和<一个href="//www.newsama.com/journals/am/2011/310762/tab5/" target="_blank">5</一个>,我们可以观察到所有用户的统计结果。可以看出,60.5%的混合排名是首选的用户查询,查询两个功能的8.5%似乎执行同样的,而其余31%的视觉排名优于混合查询。在图<一个href="//www.newsama.com/journals/am/2011/310762/fig19/" target="_blank">19</一个>,我们可以看到对应的柱状图,它显示了用户偏好(即的频率。,年代election clicks on hybrid ranking minus clicks on visual ranking) for the involved queries. We constructed the histogram by considering absolute values of the clicks and not normalized (i.e., divided by the total clicks in a query), as the actual number of clicks seems to be of more importance. For instance, in a case that a user clicks and selects only one item more from the one of the rankings, the conclusion should be that this ranking is with higher probability slightly better than the other. This can be reflected when considering the absolute difference of the clicks (e.g., 9 − 8 = 1 and 1 − 0 = 1), where the value 1 describes the user preference. However, if we normalize the metrics according to the total number of selections in a query, we could get misleading results (e.g., 11.1% and 100%, resp., for the previous example).</p><div class="floats-partial-section partial-other-wrapper" data-section="tab4"> <div class="floats-partial-content fig-content-tab4"> <table class="table-group" id="tab4"> <tbody> <tr> <td> <table class="table"> <tbody> <tr> <td class="thead-hr" colspan="4"> <hr></td> </tr> <tr class="thead"> <td align="left">总查询</td> <td align="center">更多的选择在混合</td> <td align="center">更多的选择在视觉</td> <td align="center">平等的选择</td> </tr> <tr> <td class="thead-hr" colspan="4"> <hr></td> </tr> <tr> <td align="left">200年</td> <td align="center">121年</td> <td align="center">62年</td> <td align="center">17</td> </tr> <tr class="table-tr"> <td colspan="4"> <hr class="tbody-hr"></td> </tr> </tbody> </table></td> </tr> </tbody> </table> </div> <div class="floats-partial-footer"> <div class="floats-partial-caption"> <span class="caption-text">表4</年代p一个n><一个一个r我一个-label="full view" class="caption-partial-url" href="//www.newsama.com/journals/am/2011/310762/tab4/" title="" target="_blank"></a> </div> <div class="floats-partial-caption-text"> 两两比较的混合检索功能和视觉。</div> </div> </div> <div class="floats-partial-section partial-other-wrapper" data-section="tab5"> <div class="floats-partial-content fig-content-tab5"> <table class="table-group" id="tab5"> <tbody> <tr> <td> <table class="table"> <tbody> <tr> <td class="thead-hr" colspan="3"> <hr></td> </tr> <tr class="thead"> <td align="left">总点击量</td> <td align="center">选择在混合的排名</td> <td align="center">选择在视觉的排名</td> </tr> <tr> <td class="thead-hr" colspan="3"> <hr></td> </tr> <tr> <td align="left">6509年</td> <td align="center">5276年</td> <td align="center">1233年</td> </tr> <tr class="table-tr"> <td colspan="3"> <hr class="tbody-hr"></td> </tr> </tbody> </table></td> </tr> </tbody> </table> </div> <div class="floats-partial-footer"> <div class="floats-partial-caption"> <span class="caption-text">表5</年代p一个n><一个一个r我一个-label="full view" class="caption-partial-url" href="//www.newsama.com/journals/am/2011/310762/tab5/" title="" target="_blank"></a> </div> <div class="floats-partial-caption-text"> 点击混合和视觉排名。</div> </div> </div> <div class="floats-partial-section partial-carousel-wrapper" data-section="fig19"> <div class="floats-partial-content fig-content-fig19"> <div class="partial-carousel--single-image"> <img alt="" name="310762.fig.0019" src="https://static-01.hindawi.com/articles/am/volume-2011/310762/figures/310762.fig.0019.jpg"> <br> <strong></strong> </div> </div> <div class="floats-partial-footer"> <div class="floats-partial-caption"> <span class="caption-text">图19</年代p一个n><一个一个r我一个-label="full view" class="caption-partial-url" href="//www.newsama.com/journals/am/2011/310762/fig19/" title="" target="_blank"></a> </div> <div class="floats-partial-caption-text"> <table> 直方图的点击。水平轴代表的点击数量(视觉积极混合排名和消极的),而垂直的频率。</t一个ble> </div> </div> </div> <h4 id="conclusions">10。结论</h4><p>在本文中,我们引入了新的视频搜索和隐性利益指标提出了一个新颖的方法来构造一个内容相似图基于隐式的用户交互模式的指标。此外,我们提出了一个有效的视觉特征和隐式的用户反馈相结合的方法利用支持向量机分类器。结果所示,附加值的隐式的用户反馈可以视频检索引擎,也考虑过去大量的用户交互数据很容易变得可用。从实验结果看来,利用隐式反馈能够改善视觉搜索结果在大多数的情况下,此外,它能改善系统的性能。我们可以说,利用隐式优化视觉搜索用户反馈似乎调整视觉功能语义的方式,结果与相同的语义概念的排名更高,尽管最初的视觉相似性较低。</p><p>被认为是可伸缩的方法因为它是基于良好的图形结构,提出了许多有效的索引技术。此外,行动和加权图收缩的离线,在后台更新,当新用户隐式反馈收集。因此,应用程序的方法,特别是实时(即分类任务。,混合搜索)仍可能即使在巨大的语料库。</p><p>未来的工作将包括更多的扩展方法的评价,以便更好地观察这种混合方法的行为尤其是在情况下,更多的用户参与和低质量的隐式反馈(例如,在对照实验更少)。此外,除了支持向量机分类算法的应用,例如,增强型技术和比较他们之间的利益,而就业的替代视觉特性(例如,筛选)也可以被考虑。最后,一个重要的挑战是包括信息从用户的生理反应(例如,眼球运动),并且要结合用户导航模式和视频分析信息。最近,我们的研究活动是针对这些目标。</p><h4我d="acknowledgment">承认</h4><p>这项工作是支持的项目PESCaDO (fp7 - 248594)之下,PetaMedia (fp7 - 216444)之下都由欧洲委员会资助。</p></div> <div class="xml-content"> <h4 class="references" id="references">引用</h4><ol> <li id="B1">m . s .卢:·c·Djeraba r . Jain,“基于内容的多媒体信息检索:最先进的和挑战,”<我>ACM交易多媒体计算、通信和应用程序</我>,卷2,不。1 - 2006页。<年代p一个nclass="reflinks">视图:<一个href="https://scholar.google.com/scholar_lookup?title=Content-based%20multimedia%20information%20retrieval:%20state%20of%20the%20art%20and%20challenges&author=M. S. Lew&author=N. Sebe&author=C. Djeraba&author=&author=R. Jain&publication_year=2006" target="_blank">谷歌学术搜索</一个></年代p一个n></li> <li id="B2">n·m·s·卢x周,t . s .黄,他和e . m .”艺术在图像和视频检索的状态,”<我>第二届国际会议上图像和视频检索</我>乌尔班纳,页1 - 8,生病,美国,2003年7月。<年代p一个nclass="reflinks">视图:<一个href="https://scholar.google.com/scholar_lookup?title=The%20state%20of%20the%20art%20in%20image%20and%20video%20retrieval&author=N. Sebe&author=M. S. Lew&author=X. Zhou&author=T. S. Huang&author=&author=E. M. Bakker" target="_blank">谷歌学术搜索</一个></年代p一个n></li> <li id="B3">婆罗米语m . l . Kherfi d, d . Ziou”视觉特征和语义相结合更有效的图像检索,”<我>17学报》国际会议模式识别(ICPR ' 04)</我>,卷2,页961 - 964,剑桥,英国,2004年8月。<年代p一个nclass="reflinks">视图:<一个href="https://scholar.google.com/scholar_lookup?title=Combining%20visual%20features%20with%20semantics%20for%20a%20more%20effective%20image%20retrieval&author=M. L. Kherfi&author=D. Brahmi&author=&author=D. Ziou" target="_blank">谷歌学术搜索</一个></年代p一个n></li> <li id="B4">s . f . Chang, r . Manmatha和t . s . Chua”结合文本和视听功能视频索引,”<我>《IEEE国际会议音响、演讲和信号处理(ICASSP 05)</我>,5卷,页V1005-V1008,费城,宾夕法尼亚州,美国,2005年3月。<年代p一个nclass="reflinks">视图:<一个href="https://doi.org/10.1109/ICASSP.2005.1416476">出版商的网站</一个><年代p一个nclass="sep">|</年代p一个n><一个href="https://scholar.google.com/scholar_lookup?title=Combining%20text%20and%20audio-visual%20features%20in%20video%20indexing&author=S. F. Chang&author=R. Manmatha&author=&author=T. S. Chua" target="_blank">谷歌学术搜索</一个></年代p一个n></li> <li id="B5">s . Vrochidis c . Doulaverakis A . Gounaris e . Nidelkou l . Makris i Kompatsiaris,“混合本体和visual-based文化遗产多媒体检索模型集合,”<我>国际期刊的元数据语义和本体</我>,3卷,不。3、167 - 182年,2008页。<年代p一个nclass="reflinks">视图:<一个href="https://doi.org/10.1504/IJMSO.2008.023566">出版商的网站</一个><年代p一个nclass="sep">|</年代p一个n><一个href="https://scholar.google.com/scholar_lookup?title=A%20hybrid%20ontology%20and%20visual-based%20retrieval%20model%20for%20cultural%20heritage%20multimedia%20collections&author=S. Vrochidis&author=C. Doulaverakis&author=A. Gounaris&author=E. Nidelkou&author=L. Makris&author=&author=I. Kompatsiaris&publication_year=2008" target="_blank">谷歌学术搜索</一个></年代p一个n></li> <li id="B6">c .通用杖鱼和m .烦恼”,最先进的多通道视频索引:审查,”<我>多媒体工具和应用程序</我>,25卷,不。1,5-35,2005页。<年代p一个nclass="reflinks">视图:<一个href="https://doi.org/10.1023/B:MTAP.0000046380.27575.a5">出版商的网站</一个><年代p一个nclass="sep">|</年代p一个n><一个href="https://scholar.google.com/scholar_lookup?title=Multimodal%20video%20indexing:%20a%20review%20of%20the%20state-of-the-art&author=C. G. M. Snoek &author=M. Worring&publication_year=2005" target="_blank">谷歌学术搜索</一个></年代p一个n></li> <li id="B7">f . Hopfgartner d . Vallet m . Halvey和j·穆,“搜索轨迹利用用户反馈改善视频搜索,”<我>学报》第16届ACM国际会议多媒体(毫米' 08)</我>加拿大温哥华,页339 - 348,2008年10月。<年代p一个nclass="reflinks">视图:<一个href="https://doi.org/10.1145/1459359.1459405">出版商的网站</一个><年代p一个nclass="sep">|</年代p一个n><一个href="https://scholar.google.com/scholar_lookup?title=Search%20trails%20using%20user%20feedback%20to%20improve%20video%20search&author=F. Hopfgartner&author=D. Vallet&author=M. Halvey&author=&author=J. Jose" target="_blank">谷歌学术搜索</一个></年代p一个n></li> <li id="B8">吴x周,y,科恩和t . s .黄”在基于内容的图像和视频检索相关性反馈,”<我>学报》第四届欧洲研讨会图像分析多媒体交互式服务</我>,页1 - 12,伦敦大学玛丽皇后,2003年。<年代p一个nclass="reflinks">视图:<一个href="https://scholar.google.com/scholar_lookup?title=Relevance%20feedback%20in%20content-based%20image%20and%20video%20retrieval&author=X. S. Zhou&author=Y. Wu&author=I. Cohen&author=&author=T. S. Huang" target="_blank">谷歌学术搜索</一个></年代p一个n></li> <li id="B9">g . Giacinto f . Roli,“基于实例的相关反馈图像检索,”<我>先进的神经信息处理系统</我>l . k .扫罗、y维斯和l . Bottou Eds。,17卷,页489 - 496,麻省理工学院出版社,2005年。<年代p一个nclass="reflinks">视图:<一个href="https://scholar.google.com/scholar_lookup?title=Instance-based%20relevance%20feedback%20for%20image%20retrieval&author=G. Giacinto &author=F. Roli&publication_year=2005" target="_blank">谷歌学术搜索</一个></年代p一个n></li> <li id="B10">c . Gurrin d·约翰森,a . f . Smeaton“支持在视频搜索相关性反馈,”<我>第28届欧洲会议IR研究学报》(ECIR 06年)</我>,页561 - 564,伦敦,英国,2006年。<年代p一个nclass="reflinks">视图:<一个href="https://scholar.google.com/scholar_lookup?title=Supporting%20relevance%20feedback%20in%20video%20search&author=C. Gurrin&author=D. Johansen&author=&author=A. F. Smeaton" target="_blank">谷歌学术搜索</一个></年代p一个n></li> <li id="B11">t . joachim“优化搜索引擎使用有效点击率数据,”<我>ACM SIGKDD国际会议的程序知识发现和数据挖掘(KDD ' 02)</我>,页133 - 142,加拿大阿尔伯塔省,2002年7月。<年代p一个nclass="reflinks">视图:<一个href="https://scholar.google.com/scholar_lookup?title=Optimizing%20search%20engines%20using%20clickthrough%20data&author=T. Joachims" target="_blank">谷歌学术搜索</一个></年代p一个n></li> <li id="B12">f . Radlinski和t . joachim查询链:从隐式反馈,学习排名”<我>ACM SIGKDD国际会议的程序知识发现和数据挖掘</我>芝加哥,页239 - 248年,生病,美国,2005年8月。<年代p一个nclass="reflinks">视图:<一个href="https://doi.org/10.1145/1081870.1081899">出版商的网站</一个><年代p一个nclass="sep">|</年代p一个n><一个href="https://scholar.google.com/scholar_lookup?title=Query%20chains:%20learning%20to%20rank%20from%20implicit%20feedback&author=F. Radlinski &author=T. Joachims" target="_blank">谷歌学术搜索</一个></年代p一个n></li> <li id="B13">d·m·尼科尔斯“隐式评分和过滤”<我>第五提洛岛学报》研讨会上过滤和协同过滤</我>页,31-36布达佩斯,匈牙利,1997年11月。<年代p一个nclass="reflinks">视图:<一个href="https://scholar.google.com/scholar_lookup?title=Implicit%20ratings%20and%20filtering&author=D. M. Nichols" target="_blank">谷歌学术搜索</一个></年代p一个n></li> <li id="B14">杨,t·梅x华,l·杨和m . Li Yang树群“在线视频推荐基于多通道融合和相关性反馈,”<我>学报第六届ACM国际会议上图像和视频检索(CIVR ' 07)</我>Amesterdam,页73 - 80年,荷兰,2007年7月。<年代p一个nclass="reflinks">视图:<一个href="https://doi.org/10.1145/1282280.1282290">出版商的网站</一个><年代p一个nclass="sep">|</年代p一个n><一个href="https://scholar.google.com/scholar_lookup?title=Online%20video%20recommendation%20based%20on%20multimodal%20fusion%20and%20relevance%20feedback&author=B. Yang&author=T. Mei&author=X. S. Hua&author=L. Yang&author=S. Q. Yang&author=&author=M. Li" target="_blank">谷歌学术搜索</一个></年代p一个n></li> <li id="B15">s Vrochidis i Kompatsiaris,佩特雷,“优化视觉搜索与隐式的用户反馈互动视频检索”<我>诉讼ACM国际会议上的图像和视频检索我(CIVR 10)</我>20.10年7月,页274 - 281。<年代p一个nclass="reflinks">视图:<一个href="https://doi.org/10.1145/1816041.1816082">出版商的网站</一个><年代p一个nclass="sep">|</年代p一个n><一个href="https://scholar.google.com/scholar_lookup?title=Optimizing%20visual%20search%20with%20implicit%20user%20feedback%20in%20interactive%20video%20retrieval&author=S. Vrochidis&author=I. Kompatsiaris&author=&author=I. Patras" target="_blank">谷歌学术搜索</一个></年代p一个n></li> <li id="B16">s . Vrochidis i Kompatsiaris,佩特雷,“交互式视频检索,利用隐式的用户反馈”<我>学报》第11届国际研讨会上图像分析多媒体交互式服务(WIAMIS 10)</我>,Desenzano德尔·加尔达湖,意大利,2010年4月。<年代p一个nclass="reflinks">视图:<一个href="https://scholar.google.com/scholar_lookup?title=Exploiting%20implicit%20user%20feedback%20in%20interactive%20video%20retrieval&author=S. Vrochidis&author=I. Kompatsiaris&author=&author=I. Patras" target="_blank">谷歌学术搜索</一个></年代p一个n></li> <li id="B17">h . y . Zhang傅梁z z,气,和d·冯“眼球运动作为相关性反馈的相互作用机理在基于内容的图像检索系统中,”<我>诉讼的眼动跟踪研究与应用研讨会(ETRA 10)</我>页,37-40奥斯汀,德克萨斯州,美国,2010年。<年代p一个nclass="reflinks">视图:<一个href="https://doi.org/10.1145/1743666.1743674">出版商的网站</一个><年代p一个nclass="sep">|</年代p一个n><一个href="https://scholar.google.com/scholar_lookup?title=Eye%20movement%20as%20an%20interaction%20mechanism%20for%20relevance%20feedback%20in%20a%20content-based%20image%20retrieval%20system&author=Y. Zhang&author=H. Fu&author=Z. Liang&author=Z. Chi&author=&author=D. Feng" target="_blank">谷歌学术搜索</一个></年代p一个n></li> <li id="B18">a . Yazdani j·s·李,t . Ebrahimi“隐式多媒体使用EEG信号的情感标签和大脑计算机接口,”<我>学报SIGMM研讨会上社交媒体</我>,页81 - 88,纽约,纽约,美国,2009年。<年代p一个nclass="reflinks">视图:<一个href="https://doi.org/10.1145/1631144.1631160">出版商的网站</一个><年代p一个nclass="sep">|</年代p一个n><一个href="https://scholar.google.com/scholar_lookup?title=Implicit%20emotional%20tagging%20of%20multimedia%20using%20EEG%20signals%20and%20brain%20computer%20interface&author=A. Yazdani&author=J. S. Lee&author=&author=T. Ebrahimi" target="_blank">谷歌学术搜索</一个></年代p一个n></li> <li id="B19">d·凯利和j . Teevan“隐式反馈推断用户偏好:参考书目,“<我>市立论坛</我>,32卷,不。2、2003。<年代p一个nclass="reflinks">视图:<一个href="https://scholar.google.com/scholar_lookup?title=Implicit%20feedback%20for%20inferring%20user%20preference:%20a%20bibliography&author=D. Kelly &author=J. Teevan&publication_year=2003" target="_blank">谷歌学术搜索</一个></年代p一个n></li> <li id="B20">m . Claypool p . Le, m是d·布朗,“隐性利益指标,”<我>学报》国际会议智能用户界面(IUI的01)</我>页33-40圣达菲,墨西哥人,美国,2001年。<年代p一个nclass="reflinks">视图:<一个href="https://scholar.google.com/scholar_lookup?title=Implicit%20interest%20indicators&author=M. Claypool&author=P. Le&author=M. Wased&author=&author=D. Brown" target="_blank">谷歌学术搜索</一个></年代p一个n></li> <li id="B21">r .白色,鲁斯温,和j·m·穆”的使用隐式相关性反馈在网络检索的证据,”<我>《24日BCS-IRSG欧洲IR研究讨论会:先进的信息检索</我>格拉斯哥,页93 - 109年,英国,2002年。<年代p一个nclass="reflinks">视图:<一个href="https://scholar.google.com/scholar_lookup?title=The%20use%20of%20implicit%20evidence%20for%20relevance%20feedback%20in%20web%20retrieval&author=R. White&author=I. Ruthven&author=&author=J. M. Jose" target="_blank">谷歌学术搜索</一个></年代p一个n></li> <li id="B22">f . Hopfgartner和j·穆”,评价的隐式反馈模型自适应视频检索,”<我>学报的国际多媒体信息检索</我>,页323 - 331年,巴伐利亚,德国,2007年。<年代p一个nclass="reflinks">视图:<一个href="https://doi.org/10.1145/1290082.1290127">出版商的网站</一个><年代p一个nclass="sep">|</年代p一个n><一个href="https://scholar.google.com/scholar_lookup?title=Evaluating%20the%20implicit%20feedback%20models%20for%20adaptive%20video%20retrieval&author=F. Hopfgartner &author=J. Jose" target="_blank">谷歌学术搜索</一个></年代p一个n></li> <li id="B23">d . Vallet f . Hopfgartner, j·穆”使用隐式图推荐相关视频:模拟评估,”<我>学报30年欧洲会议信息检索(ECIR 08年)</我>卷,4956<我>在计算机科学的课堂讲稿</我>,页199 - 210,格拉斯哥,苏格兰,2008年。<年代p一个nclass="reflinks">视图:<一个href="https://doi.org/10.1007/978-3-540-78646-7_20">出版商的网站</一个><年代p一个nclass="sep">|</年代p一个n><一个href="https://scholar.google.com/scholar_lookup?title=Use%20of%20implicit%20graph%20for%20recommending%20relevant%20videos:%20a%20simulated%20evaluation&author=D. Vallet&author=F. Hopfgartner&author=&author=J. Jose" target="_blank">谷歌学术搜索</一个></年代p一个n></li> <li id="B24">约阿希姆,“在线性时间内线性支持向量机训练,”<我>第12届ACM SIGKDD国际会议程序知识发现和数据挖掘(KDD ' 06)</我>,页217 - 226年,费城,宾夕法尼亚州,美国,2006年8月。<年代p一个nclass="reflinks">视图:<一个href="https://scholar.google.com/scholar_lookup?title=Training%20linear%20SVMs%20in%20linear%20time&author=T. Joachims" target="_blank">谷歌学术搜索</一个></年代p一个n></li> <li id="B25">t·约阿希姆,“多元支持向量方法性能的措施,”<我>学报》第二十二届国际会议上机器学习(ICML ' 05)</我>波恩,页377 - 384年,德国,2005年8月。<年代p一个nclass="reflinks">视图:<一个href="https://doi.org/10.1145/1102351.1102399">出版商的网站</一个><年代p一个nclass="sep">|</年代p一个n><一个href="https://scholar.google.com/scholar_lookup?title=A%20support%20vector%20method%20for%20multivariate%20performance%20measures&author=T. Joachims" target="_blank">谷歌学术搜索</一个></年代p一个n></li> <li id="B26">A . Hanjalic r . l . Lagendijk, j . Biemond”新方法基于关键帧的视频内容表示,”<我>图像数据库和多媒体搜索</我>一个。Smeulders和r . Jain, Eds。,pp。97- - - - - -107,World Scientific, 1997.<年代p一个nclass="reflinks">视图:<一个href="https://scholar.google.com/scholar_lookup?title=A%20new%20method%20for%20key%20frame%20based%20video%20content%20representation&author=A. Hanjalic&author=R. L. Lagendijk&author=&author=J. Biemond&publication_year=1997" target="_blank">谷歌学术搜索</一个></年代p一个n></li> <li id="B27">美国美国纳希,Damnjanovic, b . Mansencal et al .,“冲总结COST292实验框架的任务,”<我>学报TRECVID车间</我>美国马里兰州盖瑟斯堡,页40-44,,,2008。<年代p一个nclass="reflinks">视图:<一个href="https://doi.org/10.1145/1463563.1463569">出版商的网站</一个><年代p一个nclass="sep">|</年代p一个n><一个href="https://scholar.google.com/scholar_lookup?title=The%20COST292%20experimental%20framework%20for%20rushes%20summarization%20task&author=S. U. Naci&author=U. Damnjanovic&author=B. Mansencal et al." target="_blank">谷歌学术搜索</一个></年代p一个n></li> <li id="B28">Kinosearch搜索引擎库,<一个t一个rget="_blank" href="http://www.rectangular.com/kinosearch/">http://www.rectangular.com/kinosearch/</一个>。<年代p一个nclass="reflinks"></span></li> <li id="B29">m·f·波特,”一个后缀剥离算法”,<我>程序</我>,14卷,第137 - 130页,1980年。<年代p一个nclass="reflinks">视图:<一个href="https://scholar.google.com/scholar_lookup?title=An%20algorithm%20for%20suffix%20stripping&author=M. F. Porter&publication_year=1980" target="_blank">谷歌学术搜索</一个></年代p一个n></li> <li id="B30">s e·罗伯逊和k·s·琼斯,“简单,证明文本检索方法,”技术。众议员ucam - cl - tr - 356年,剑桥大学计算机实验室,剑桥,英国,1994年。<年代p一个nclass="reflinks">视图:<一个href="https://scholar.google.com/scholar_lookup?title=Simple,%20proven%20approaches%20to%20text%20retrieval&author=S. E. Robertson &author=K. S. Jones&publication_year=1994" target="_blank">谷歌学术搜索</一个></年代p一个n></li> <li id="B31">MPEG-7 XM的软件,<一个t一个rget="_blank" href="http://www.lis.ei.tum.de/research/bv/topics/mmdb/e_mpeg7.html">http://www.lis.ei.tum.de/research/bv/topics/mmdb/e_mpeg7.html</一个>。<年代p一个nclass="reflinks"></span></li> <li id="B32">a .格特曼“r - tree:空间搜索的动态索引结构,”<我>ACM国际会议的程序管理和数据(SIGMOD 84)</我>,页47-57,纽约,纽约,美国,1984年。<年代p一个nclass="reflinks">视图:<一个href="https://scholar.google.com/scholar_lookup?title=R-trees:%20a%20dynamic%20index%20structure%20for%20spatial%20searching&author=A. Guttman" target="_blank">谷歌学术搜索</一个></年代p一个n></li> <li id="B33">s . Vrochidis p .国王,l . Makris et al .,“MKLab交互式视频检索系统”<我>诉讼ACM国际会议上的图像和视频检索(CIVR 08年)</我>尼亚加拉大瀑布,页563 - 563年,加拿大,2008。<年代p一个nclass="reflinks">视图:<一个href="https://doi.org/10.1145/1646396.1646459">出版商的网站</一个><年代p一个nclass="sep">|</年代p一个n><一个href="https://scholar.google.com/scholar_lookup?title=MKLab%20interactive%20video%20retrieval%20system&author=S. Vrochidis&author=P. King&author=L. Makris et al." target="_blank">谷歌学术搜索</一个></年代p一个n></li> <li id="B34">e·w·迪杰斯特拉”,注意在连接图,两个问题”<我>Numerische Mathematik</我>,1卷,不。1,第271 - 269页,1959。<年代p一个nclass="reflinks">视图:<一个href="https://doi.org/10.1007/BF01386390">出版商的网站</一个><年代p一个nclass="sep">|</年代p一个n><一个href="https://scholar.google.com/scholar_lookup?title=A%20note%20on%20two%20problems%20in%20connexion%20with%20graphs&author=E. W. Dijkstra&publication_year=1959" target="_blank">谷歌学术搜索</一个></年代p一个n></li> <li id="B35">r·w·弗洛伊德“97年算法:最短路径,”<我>ACM的通信</我>,5卷,不。6,345年,页1962。<年代p一个nclass="reflinks">视图:<一个href="https://scholar.google.com/scholar_lookup?title=Algorithm%2097:%20shortest%20path&author=R. W. Floyd&publication_year=1962" target="_blank">谷歌学术搜索</一个></年代p一个n></li> </ol> </div> <div class="xml-content"> <h4 class="copyright" id="copyright">版权</h4><p>版权©2011年塞浦路斯Vrochidis等。这是一个开放的分布式下文章<一个rel="license" href="http://creativecommons.org/licenses/by/3.0/">知识共享归属许可</一个>,它允许无限制的使用、分配和复制在任何媒介,提供最初的工作是正确引用。</p></div> </div> </article> </div> </div> </div> <div class="biblio__Wrapper-sc-1nou1lz-0 eyHfWh journal_article__BiblioIntegratorWidget-sc-12fk6zy-5 dbAANp article_widget biblioWrapper hide"> <hr> <h4 class="biblio__H4-sc-1nou1lz-2 becpDt hrTitleRef">更多相关文章</h4><div id="biblio_widget" class="biblio__BiblioWrapper-sc-1nou1lz-1 hFGNjP"></div> </div> </div> </div> <div class="threeColumn__RightWrapper-sc-1bf4078-2 jzTMLK rightWrapper" id="rightSectionMenu"> <div> <div class="journal_article__ArticleStickyTopToolbar-sc-12fk6zy-0 gbrQxi article_sticky_top_toolbar"> <div class="sc-cBdUnI hRnXob article_top_toolbar"> <a class="sc-htpNat bUhGXt link sc-jtRfpW dIjoCg sc-exkUMo hymroZ pdf_desktop button__group" href="//www.newsama.com/downloads/journals/am/2011/310762.pdf" aria-label="" target="_blank" id="toolbar__dcNavSmall"><span><img alt="" class="sc-EHOje jOLhQl sc-kTUwUJ eujois" title="" role="presentation" src="https://static-01.hindawi.com/next_assets/wyT4heYybLmt1UEJexPE9/_next/static/node_modules/hindawi-ui/src/icon/svg/pdf.svg" height="24"><span class="text">PDF</年代p一个n></年代p一个n><span class="sc-htoDjs bqeMOy sc-dqBHgY cQgXqt fileArrow icon-colored _3P4iUn1NDM5G77nWCamJ4l icon-themeable" title="" font-size="20"></span></a> <a class="sc-htpNat bUhGXt link sc-jtRfpW dIjoCg sc-kcDeIU cbcyj ref_desktop" href="//www.newsama.com/journals/am/2011/310762.ris" aria-label="" target="" id="toolbar__dcNavLarge"><span><img alt="" class="sc-EHOje jOLhQl sc-kTUwUJ eujois" title="" role="presentation" src="data:image/svg+xml;base64,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" height="24"><span class="text">下载引用</年代p一个n></年代p一个n><span class="sc-htoDjs bqeMOy sc-dqBHgY cQgXqt fileArrow icon-colored _3P4iUn1NDM5G77nWCamJ4l icon-themeable" title="" font-size="20"></span></a> <a class="sc-htpNat bUhGXt link sc-jtRfpW dIjoCg sc-BngTV hmDWcT ref_desktop" href="//www.newsama.com/journals/am/2011/310762.ris" aria-label="" target="" id="toolbar__dcNavSmall"><span><img alt="" class="sc-EHOje jOLhQl sc-kTUwUJ eujois" title="" role="presentation" src="data:image/svg+xml;base64,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" height="24"><span class="text">引用</年代p一个n></年代p一个n><span class="sc-htoDjs bqeMOy sc-dqBHgY cQgXqt fileArrow icon-colored _3P4iUn1NDM5G77nWCamJ4l icon-themeable" title="" font-size="20"></span></a> </div> </div> <div class="journal_article__ArticleStickyBottomToolbar-sc-12fk6zy-1 isgRZA article_sticky_bottom_toolbar"> <div class="sc-cTjmhe ihmyEq article_bottom_toolbar"> <div class="sc-cugefK itFKcQ accordion" property="background-color" color="#D8D8D8" id="toolbar__dofNav"> <span><img alt="" class="sc-EHOje jOLhQl sc-bfYoXt cGuwDn" title="" role="presentation" src="data:image/svg+xml;base64,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" height="24"><span class="sc-dRFtgE jrsDZP">下载其他格式</年代p一个n><年代p一个nclass="sc-gkFcWv gWTCXQ">更多的</年代p一个n></年代p一个n> <span class="sc-htoDjs bMdQcY sc-gbOuXE jGPvPL icon-colored _1CwL9R82TPl2HxDleeYGDR icon-themeable" title="" font-size="11"></span> </div> <div class="sc-fnwBNb etzuBr"> <a class="sc-htpNat bUhGXt link sc-iNhVCk gWgCNb open_panel" href="//www.newsama.com/downloads/journals/am/2011/310762.xml" aria-label="XML" property="background-color" color="#D8D8D8"> <div class="sc-eAKXzc WHkfb"> XML</div><img alt="" class="sc-EHOje jOLhQl sc-bfYoXt cGuwDn" title="" role="presentation" src="data:image/svg+xml;base64,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" height="24"></a> </div> <a class="sc-htpNat bUhGXt link sc-imABML ibWUzs" href="//www.newsama.com/reprints/310762/confirm" aria-label="Order Printed Copies" target="_blank" property="background-color" color="#D8D8D8" id="toolbar__opcNav"><span><img alt="" class="sc-EHOje jOLhQl sc-bfYoXt cGuwDn" title="" role="presentation" src="data:image/svg+xml;base64,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" height="24"><span class="sc-dRFtgE jrsDZP">订单打印副本</年代p一个n><年代p一个nclass="sc-gkFcWv gWTCXQ">订单</年代p一个n></年代p一个n><span class="sc-htoDjs kNoiZu sc-gbOuXE jGPvPL right__icon icon-colored _32KP5uexH-Bgs7Kj8Gk1Tz icon-themeable" title="" font-size="16"></span></a> <div class="sc-hBbWxd kdRRST"> <div class="sc-dyGzUR cHatnS"> <div> 的观点<年代p一个n>2090年年</年代p一个n></div> <div> 下载<年代p一个n>772年年</年代p一个n></div> </div> <div> <div class="sc-drKuOJ kxdGAZ"> <span class="__dimensions_badge_embed__" data-doi="10.1155/2011/310762" data-style="small_circle"></span>引用</div> </div> </div> <nav class="sc-bRBYWo fXBKFa sc-hUfwpO jHPgUP" id="toolbar__iconNav"> <h3 class="sc-Rmtcm hNPrIr">分享文章</h3><ul class="sc-hzDkRC bnbfTS"> <li class="iconBorder"><a class="sc-htpNat bUhGXt link" href="https://twitter.com/share?url=https://doi.org/10.1155/2011/310762&via=Hindawi" aria-label="Sharing on Twitter" rel="noopener noreferrer" target="_blank"><img alt="" class="sc-EHOje jOLhQl socialTwitter" title="" role="presentation" src="data:image/svg+xml;base64,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" height="24"></a></li> <li class="iconBorder"><a class="sc-htpNat bUhGXt link" href="https://www.linkedin.com/shareArticle/?url=https://doi.org/10.1155/2011/310762" aria-label="Sharing on Linkedin" rel="noopener noreferrer" target="_blank"><img alt="" class="sc-EHOje jOLhQl socialLinkedin" title="" role="presentation" src="https://static-01.hindawi.com/next_assets/wyT4heYybLmt1UEJexPE9/_next/static/node_modules/hindawi-ui/src/icon/svg/socialLinkedin.svg" height="24"></a></li> <li class="iconBorder"><a class="sc-htpNat bUhGXt link" href="https://www.facebook.com/sharer/sharer.php?u=https://doi.org/10.1155/2011/310762" aria-label="Sharing on Facebook" rel="noopener noreferrer" target="_blank"><img alt="" class="sc-EHOje jOLhQl socialFacebook" title="" role="presentation" src="https://static-01.hindawi.com/next_assets/wyT4heYybLmt1UEJexPE9/_next/static/node_modules/hindawi-ui/src/icon/svg/socialFacebook-HINDAWI.svg" height="24"></a></li> <li class="iconBorder"> <div class="wechat-container"> <img alt="" class="sc-EHOje jOLhQl socialWechat" title="" role="presentation" src="data:image/svg+xml;base64,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" height="24"> <div class="addthis_inline_share_toolbox"></div> </div></li> <li><a class="sc-htpNat bUhGXt link" href="mailto:?subject=Utilizing Implicit User Feedback to Improve Interactive Video Retrieval&body=https://doi.org/10.1155/2011/310762" aria-label="Sharing on Mail" target="_blank"><img alt="" class="sc-EHOje jOLhQl socialMail" title="" role="presentation" src="data:image/svg+xml;base64,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" height="24"></a></li> </ul> </nav> </div> </div> <div class="biblio__Wrapper-sc-1nou1lz-0 cBTaXT journal_article__BiblioIntegratorSidebar-sc-12fk6zy-4 cYsVdk biblioWrapperRight biblioWrapper hide"> <hr> <h4 class="biblio__H4-sc-1nou1lz-2 becpDt hrTitleRef">相关文章</h4><div id="biblio_sidebar" class="biblio__BiblioWrapper-sc-1nou1lz-1 hFGNjP"></div> </div> </div> </div> </div> </div> </main> <footer class="sc-jKJlTe inBNwS"> <section class="sc-eNQAEJ dkutRV"> <div class="sc-bZQynM bNmdOy E_HUmlU2Hj3PpkDmWzKPd primary__footer footer__bodypadding"> <div class="sc-hMqMXs hlGpbO mobile"> 遵循我们:<n一个v role="navigation" class="sc-kkGfuU filPdg"> <a class="sc-htpNat bUhGXt link sc-kEYyzF ecHxGW" href="https://twitter.com/hindawi" aria-label="Twitter" rel="noopener noreferrer" target="_blank"><img alt="" class="sc-EHOje TsaZF" title="" role="presentation" src="data:image/svg+xml;base64,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" height="40"></a> <a class="sc-htpNat bUhGXt link sc-kEYyzF ecHxGW" href="https://www.linkedin.com/company/hindawilimited/" aria-label="LinkedIn" rel="noopener noreferrer" target="_blank"><img alt="" class="sc-EHOje TsaZF" title="" role="presentation" src="data:image/svg+xml;base64,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" height="40"></a> <a class="sc-htpNat bUhGXt link sc-kEYyzF ecHxGW" href="https://www.youtube.com/channel/UCDIK_O7FATPEkhad_sQVaSg" aria-label="Youtube" rel="noopener noreferrer" target="_blank"><img alt="" class="sc-EHOje TsaZF" title="" role="presentation" src="data:image/svg+xml;base64,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" height="40"></a> <a class="sc-htpNat bUhGXt link sc-kEYyzF ecHxGW" href="https://www.facebook.com/HindawiLimited/" aria-label="Facebook" rel="noopener noreferrer" target="_blank"><img alt="" class="sc-EHOje TsaZF" title="" role="presentation" src="data:image/svg+xml;base64,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" height="40"></a> </nav> </div> <nav class="E_HUmlU2Hj3PpkDmWzKPd _3wm8FgzqRuRWouJG31YgKx" role="navigation"> <ul class="_2k5c8zEhMT-84kRD4oDgAz _1SoDWGr_FVsil9klRfjnhb"> <li><a class="sc-htpNat bUhGXt link about_us" href="//www.newsama.com/about/" aria-label="About us">关于我们</一个></li> <li><a class="sc-htpNat bUhGXt link contact_us" href="//www.newsama.com/about/contact/" aria-label="Contact us">联系我们</一个></li> <li><a class="sc-htpNat bUhGXt link partnerships" href="//www.newsama.com/publishing-partnerships/" aria-label="Partnerships">合作伙伴关系</一个></li> <li><a class="sc-htpNat bUhGXt link blog" href="//www.newsama.com/blog/" aria-label="Blog">博客</一个></li> </ul> <ul class="_2k5c8zEhMT-84kRD4oDgAz _1SoDWGr_FVsil9klRfjnhb"> <li><a class="sc-htpNat bUhGXt link journals" href="//www.newsama.com/journals/" aria-label="Journals">期刊</一个></li> <li><a class="sc-htpNat bUhGXt link article_processing_charges" href="//www.newsama.com/publish-research/authors/article-processing-charges/" aria-label="Article Processing Charges">文章处理费用</一个></li> <li><a class="sc-htpNat bUhGXt link print_editions" href="//www.newsama.com/print-editions/" aria-label="Print editions">印刷版</一个></li> </ul> <ul class="_2k5c8zEhMT-84kRD4oDgAz _1SoDWGr_FVsil9klRfjnhb"> <li><a class="sc-htpNat bUhGXt link authors" href="//www.newsama.com/publish-research/authors/" aria-label="Authors">作者</一个></li> <li><a class="sc-htpNat bUhGXt link editors" href="//www.newsama.com/publish-research/editors/" aria-label="Editors">编辑器</一个></li> <li><a class="sc-htpNat bUhGXt link reviewers" href="//www.newsama.com/publish-research/reviewers/" aria-label="Reviewers">评论家</一个></li> </ul> <ul class="_2k5c8zEhMT-84kRD4oDgAz _1SoDWGr_FVsil9klRfjnhb"> <li><a class="sc-htpNat bUhGXt link hindawi_xml_corpus" href="//www.newsama.com/hindawi-xml-corpus/" aria-label="Hindawi XML Corpus">雷电竞APP</a></li> <li><a class="sc-htpNat bUhGXt link open_archives_initiative" href="//www.newsama.com/open-archives-initiative/" aria-label="Open Archives Initiative">开放档案计划</一个></li> <li><a class="sc-htpNat bUhGXt link fraud_prevention" href="//www.newsama.com/preventing-and-protecting-against-fraud/" aria-label="Fraud prevention">欺诈防范</一个></li> </ul> </nav> <div class="sc-hMqMXs hlGpbO desktop"> 遵循我们:<n一个v role="navigation" class="sc-kkGfuU filPdg"> <a class="sc-htpNat bUhGXt link sc-kEYyzF ecHxGW" href="https://twitter.com/hindawi" aria-label="Twitter" rel="noopener noreferrer" target="_blank"><img alt="" class="sc-EHOje TsaZF" title="" role="presentation" src="data:image/svg+xml;base64,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" height="40"></a> <a class="sc-htpNat bUhGXt link sc-kEYyzF ecHxGW" href="https://www.linkedin.com/company/hindawilimited/" aria-label="LinkedIn" rel="noopener noreferrer" target="_blank"><img alt="" class="sc-EHOje TsaZF" title="" role="presentation" src="data:image/svg+xml;base64,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" height="40"></a> <a class="sc-htpNat bUhGXt link sc-kEYyzF ecHxGW" href="https://www.youtube.com/channel/UCDIK_O7FATPEkhad_sQVaSg" aria-label="Youtube" rel="noopener noreferrer" target="_blank"><img alt="" class="sc-EHOje TsaZF" title="" role="presentation" src="data:image/svg+xml;base64,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" height="40"></a> <a class="sc-htpNat bUhGXt link sc-kEYyzF ecHxGW" href="https://www.facebook.com/HindawiLimited/" aria-label="Facebook" rel="noopener noreferrer" target="_blank"><img alt="" class="sc-EHOje TsaZF" title="" role="presentation" src="data:image/svg+xml;base64,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" height="40"></a> </nav> </div> </div> </section> <div class="sc-bZQynM bNmdOy"> <section class="sc-iAyFgw gCiATD"> <a class="sc-htpNat bUhGXt link" href="//www.newsama.com/privacy/" aria-label="Privacy Policy">隐私政策 |</一个><一个class="sc-htpNat bUhGXt link" href="//www.newsama.com/terms/" aria-label="Terms of Service">服务条款 |</一个><一个class="sc-htpNat bUhGXt link" href="//www.newsama.com/responsible-disclosure-policy/" aria-label="Responsible Disclosure Policy">负责信息披露政策 |</一个><一个class="sc-htpNat bUhGXt link" href="//www.newsama.com/cookie-policy/" aria-label="Cookie Policy">饼干的政策 |</一个><一个class="sc-htpNat bUhGXt link" href="//www.newsama.com/copyright/" aria-label="Copyright">版权 |</一个><一个class="sc-htpNat bUhGXt link" href="//www.newsama.com/modern-slavery-statement/" aria-label="Modern slavery statement">现代奴隶制的声明</a> </section> </div> </footer> <div class="sc-OxbzP eEFYIL advWrapper hideBanner"> <span><p>文章奖:2020年杰出的研究贡献,选择由我们的首席编辑。<一个href="//www.newsama.com/article-year-award/" rel="noopener noreferrer" target="_blank">获奖的文章阅读</一个>。</p></年代p一个n><button class="sc-bYnzgO dsXEHZ"><img alt="" class="sc-EHOje iiusBg sc-cPuPxo iXKuYd" title="" role="presentation" src="data:image/svg+xml;base64,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" height="16"></button> </div> </div> </div> <noscript> <iframe src="https://www.newsama.com/ns.html?id=GTM-MQ4MGW" height="0" width="0" style="display:none;visibility:hidden"></iframe> </noscript> </div> </body> </html>