基于自适应在线双词主题模型的应用程序评论新兴主题识别（英文）-豆柴文库

您所在位置：网站首页 / 基于自适应在线双词主题模型的应用程序评论新兴主题识别（英文）.docx / 文档详情

在线预览结束，喜欢就下载吧，查找使用更方便

5 金币

下载文档

/ 2

下载提示

如果您无法下载资料，请参考说明：

1、部分资料下载需要金币，请确保您的账户上有足够的金币

2、已购买过的文档，再次下载不重复扣费

3、资料包下载后请先用软件解压，在使用对应软件打开

文本预览

基于自适应在线双词主题模型的应用程序评论新兴主题识别（英文）
Title:EmergingTopicIdentificationinApplicationReviewsusingAdaptiveOnlineBitermTopicModel
Abstract:
Withtherapidgrowthofonlineapplications,userreviewshavebecomeanimportantsourceofinformationfordecision-making.Identifyingtheemergingtopicsinapplicationreviewscanhelpdevelopersunderstandusers'needsandimprovetheirproductsaccordingly.Inthispaper,weproposeamethodbasedontheAdaptiveOnlineBitermTopicModel(AOBTM)toidentifyemergingtopicsinapplicationreviews.
1.Introduction
Theonlineuserreviewsofapplicationsareavaluableresourceforunderstandinguserpreferences,identifyingemergingtrends,andimprovingappquality.Traditionaltopicmodelingapproaches,suchasLatentDirichletAllocation(LDA),havebeenwidelyusedforanalyzingtextdata.However,theyarenotwell-suitedforonlineapplicationsduetotheirstaticnatureandinabilitytoadapttonewtopics.
2.RelatedWork
Previousresearchhasexploredvariousapproachestoidentifyemergingtopics.Somestudieshavefocusedonfrequency-basedmethods,suchasclusteringorstatisticalanalysisofkeywords.Othershaveutilizedtopicmodelingtechniques,suchasLDA.However,thesemethodshavelimitationsincapturingthedynamicnatureofemergingtopics.
3.AdaptiveOnlineBitermTopicModel
TheAdaptiveOnlineBitermTopicModel(AOBTM)isanextensionoftheOnlineBitermTopicModel(OBTM)thataddressestheissueoftopicevolution.AOBTMadaptivelylearnsthetopicdistributionandtopic-worddistributionwhiletakingintoaccountthedynamicchangesinthecorpus.Itallowsfortheidentificationofemergingtopicsinreal-timebyadaptingtonewdocuments.
4.ExperimentalSetup
ToevaluatetheeffectivenessoftheproposedAOBTMmethod,wecollectedadatasetofapplicationreviewsfrommultipleonlineplatforms.WecomparedtheresultsofAOBTMwithLDAandotherstate-of-the-arttopicmodelingapproaches.Evaluationmetricssuchascoherencescoreandtopicdistinctivenesswereusedtoassessthequalityoftheidentifiedtopics.
5.ResultsandAnalysis
TheexperimentalresultsdemonstratedthatAOBTMsurpassesothermethodsinaccuratelyidentifyingemergingtopicsinapplicationreviews.Theadaptivenature