Text document clustering based on neighbors

Text document clustering based on neighbors

ID:39756624

大小:803.63 KB

页数:18页

时间:2019-07-10

Text document clustering based on neighbors_第1页
Text document clustering based on neighbors_第2页
Text document clustering based on neighbors_第3页
Text document clustering based on neighbors_第4页
Text document clustering based on neighbors_第5页
资源描述:

《Text document clustering based on neighbors》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库

1、Data&KnowledgeEngineering68(2009)1271–1288ContentslistsavailableatScienceDirectData&KnowledgeEngineeringjournalhomepage:www.elsevier.com/locate/datakTextdocumentclusteringbasedonneighborsabc,*CongnanLuo,YanjunLi,SoonM.ChungaTeradataCorporation,SanDiego,CA92127,USAbDepartmentofComputera

2、ndInformationScience,FordhamUniversity,Bronx,NY10458,USAcDepartmentofComputerScienceandEngineering,WrightStateUniversity,Dayton,OH45435,USAarticleinfoabstractArticlehistory:Clusteringisaverypowerfuldataminingtechniquefortopicdiscoveryfromtextdocu-Received17February2008ments.Thepartitio

3、nalclusteringalgorithms,suchasthefamilyofk-means,arereportedReceivedinrevisedform20June2009performingwellondocumentclustering.Theytreattheclusteringproblemasanoptimi-Accepted22June2009zationprocessofgroupingdocumentsintokclusterssothataparticularcriterionfunctionAvailableonline1July200

4、9isminimizedormaximized.Usually,thecosinefunctionisusedtomeasurethesimilaritybetweentwodocumentsinthecriterionfunction,butitmaynotworkwellwhentheclus-Keywords:tersarenotwellseparated.Tosolvethisproblem,weappliedtheconceptsofneighborsandDocumentclusteringlink,introducedin[S.Guha,R.Rasto

5、gi,K.Shim,ROCK:arobustclusteringalgorithmforcat-Textminingk-meansegoricalattributes,InformationSystems25(5)(2000)345–366],todocumentclustering.IfBisectingk-meanstwodocumentsaresimilarenough,theyareconsideredasneighborsofeachother.AndthePerformanceanalysislinkbetweentwodocumentsrepresen

6、tsthenumberoftheircommonneighbors.Insteadofjustconsideringthepairwisesimilarity,theneighborsandlinkinvolvetheglobalinforma-tionintothemeasurementoftheclosenessoftwodocuments.Inthispaper,weproposetousetheneighborsandlinkforthefamilyofk-meansalgorithmsinthreeaspects:anewmethodtoselectini

7、tialclustercentroidsbasedontheranksofcandidatedocuments;anewsimilaritymeasurewhichusesacombinationofthecosineandlinkfunctions;andanewheuristicfunctionforselectingaclustertosplitbasedontheneighborsoftheclustercentroids.Ourexperimentalresultsonreal-lifedatasetsdemonstratedthatourpropos

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。