a novel traffic analysis for identifying search fields in the long tail of web sites -2010

a novel traffic analysis for identifying search fields in the long tail of web sites -2010

ID:3899043

大小:393.88 KB

页数:10页

时间:2017-11-25

a novel traffic analysis for identifying search fields in the long tail of web sites -2010_第1页
a novel traffic analysis for identifying search fields in the long tail of web sites -2010_第2页
a novel traffic analysis for identifying search fields in the long tail of web sites -2010_第3页
a novel traffic analysis for identifying search fields in the long tail of web sites -2010_第4页
a novel traffic analysis for identifying search fields in the long tail of web sites -2010_第5页
资源描述:

《a novel traffic analysis for identifying search fields in the long tail of web sites -2010》由会员上传分享,免费在线阅读,更多相关内容在行业资料-天天文库

1、ANovelTrafficAnalysisforIdentifyingSearchFieldsintheLongTailofWebSitesGeorgeForman,EvanKirshenbaum,ShyamsundarRajaramHPLaboratoriesHPL-2010-27Keyword(s):webdatamining,clickstreamanalysis,machinelearningclassification,activelearningAbstract:Usingaclickstreamsample

2、of2billionURLsfrommanythousandvolunteerWebusers,wewishtoanalyzetypicalusageofkeywordsearchesacrosstheWeb.Inordertodothis,weneedtobeabletodeterminewhetheragivenURLrepresentsakeywordsearchand,ifso,whichfieldcontainsthequery.Althoughitiseasytorecognize'q'asthequeryf

3、ieldin'http://www.google.com/search?hl=en&q=music',wemustdothisautomaticallyforthelongtailofdiversewebsites.Thisproblemisthefocusofthispaper.Sincethenames,typesandnumberoffieldsdifferacrosssites,thisdoesnotconformtotraditionaltextclassificationortomulti-classprob

4、lemformulations.Theproblemalsoexhibitshighlynon-uniformimportanceacrosswebsites,sincetrafficfollowsaZipfdistribution.Wedevelopedasolutionbasedonmanuallyidentifyingthequeryfieldsonthemostpopularsites,followedbyanadaptationofmachinelearningfortherest.Itinvolvesanin

5、terestingcase-instancesstructure:labelingeachwebsitecaseusuallyinvolvesselectingatmostoneofthefieldinstancesaspositive,basedonseeingsamplefieldvalues.Thisproblemstructureandsoftconstraint-whichwebelievehasbroaderapplicability-canbeusedtogreatlyreducethemanuallabe

6、lingeffort.WeemployedactivelearningandjudiciousGUIpresentationtoefficientlytrainaclassifierwithaccuracyestimatedat96%,beatingseveralbaselinealternatives.ExternalPostingDate:February21,2010[Fulltext]ApprovedforExternalPublicationInternalPostingDate:February21,2010

7、[Fulltext]Tobepublishedandpresentedatthe19thInternationalWorldWideWebConference(WWW2010).Raleigh,NC.April26-30.2010.http://www2010.org©Copyrightthe19thInternationalWorldWideWebConference(WWW2010).ANovelTrafficAnalysisforIdentifyingSearchFieldsintheLongTailofWebSit

8、esGeorgeFormanEvanKirshenbaumShyamsundarRajaramghforman@hpl.hp.comevan.kirshenbaum@hp.comshyam.rajaram@hp.comHPLabs1501PageMillRd.PaloAlto,CA,94304USAABSTRACT1

当前文档最多预览五页,下载文档查看全文

此文档下载收益归作者所有

当前文档最多预览五页,下载文档查看全文
温馨提示:
1. 部分包含数学公式或PPT动画的文件,查看预览时可能会显示错乱或异常,文件下载后无此问题,请放心下载。
2. 本文档由用户上传,版权归属用户,天天文库负责整理代发布。如果您对本文档版权有争议请及时联系客服。
3. 下载前请仔细阅读文档内容,确认文档内容符合您的需求后进行下载,若出现内容与标题不符可向本站投诉处理。
4. 下载文档时可能由于网络波动等原因无法下载或下载错误,付费完成后未能成功下载的用户请联系客服处理。