基于MB-LDA模型微博主题挖掘.pdfVIP

计算机研究与发展 ISSN 11—1777／TP 1000—12391CN of Researchand JournalComputer Development 基于MB-LDA模型的微博主题挖掘 张晨逸1 孙建伶1 丁轶群2 1(浙江大学计算机科学与技术学院杭州310027) 2(浙江大学工业技术研究院杭州310027) (zhangehenyi．zju@gmail．corn) Model for BasedonMB—LDA TopicMiningMicroblog and Jianlin91Ding ZhangChenyil．Sun Yiqun2 Scienceand 310027) 1(CollegeofComputer Technology，ZhejiangUniversity，Hangzhou |(IndustrialResearchInstitute 310027) Technology ofZhejiangUniversity，Hangzhou AbstractAs more likeTwitterhavebecomeinformation microbloggrowspopular，services providers ona web workon focusedmoreonitsuser and scale．Early microblog relationshipcommunity thevalueofcontent．Sotheresearchon a structure，without considering microblogrequireschange from user’S toitscontent traditionaltext solely relationshipanalysis mining．Although mining havebeenstudied is methods well，no for data，whichcontain algorithmdesignedspeciallymicroblog onsocialnetworkbesides structuredinformation text．Inthis a novel plain paper，wepropose based is model onLDA，calledMB-LDA，whichsuitabletomodelthe probabilisticgenerative dataandtakesbothcontactrelationanddocumentrelationintoconsiderationto microblog