欢迎来到天天文库
浏览记录
ID:34580149
大小:11.06 MB
页数:60页
时间:2019-03-08
《基于微博的情感分析关键技术研究》由会员上传分享,免费在线阅读,更多相关内容在教育资源-天天文库。
1、万方数据II万方数据AbstractResearchonSentimentAnalysisandOpinionMiningofWeiboCandidate:JuYu,Supervisor:Prof.QiaoWANGSchoolofInformationScienceandEngineering,SoutheastUniversity,ChinaAftersteppingintotheeraofmobileinternet,peoplearebecomingmorewillingtosharetheiropinionsandlifeexperiencesw
2、ithothersthroughinternet.Meanwhile,itresultsinanexplosiveaccumulationofsubjectiveinformationonsocialnetworkslikeWeibo.Andinthiscontext,thesentimentanalysisofWeibokeepsbothgreatsocialandcommercialvalues;asforgovernors,itCanfacilitatetheirunderstandingtowardspublicopinions;whilefor
3、merchants,ithelpsthemtofollowupconsumers’attitudes.HereIaimtodevelopasetofmethodsforsentimentanalysisandopinionminingofWeibo.Myworkmainlycomprisesfourparts,i.e.unknownwordsrecognition,Weibocorpuspre-processing,sentimentlexiconsextensionandalgorithmsforsentimentanalysis.Throughthe
4、work,wehavesignificantlyimprovedtherecognitionrateofthewordsegmentationsystem.Andasentim、entlexiconforWeibowasbuiltbyus.WehavealsoconstructedintelligentminingmethodforthesentimentpolarityanalysisofWeibo.Finally,weachievedtobuildupaproto—systemforthesentimentanalysisandopinionmini
5、ngofhottopicsonWeibo.Andthroughitwecangetthesentimentpolarityoftopics,aswellasacquiretherealopinionsofWeibousers.Inthispaper,wefirstgiveabriefintroductiontothecurrentstudyofsentimentanalysisandopinionmining,anditssocialandcommercialvalue.Afterthedissectionofthelinguisticcharacter
6、isticofcontentonWeibo,weevaluatedtheeffectivenessofdatapre—processingmethodsandalgorithmsforsentimentanalysisofWeibo.Ondatapre—processing,weprovideanalgorithmforunknownwordsrecognitionwhichbasedonco-occurrencefrequency,anditcaneffectivelyrecognizeunknownwordsfromWeiboandimproveth
7、eperformanceofwordsegmentationsystem.Thesentimentlexiconsexpansionisstartedbypre—processingofWeibocorpusbasedonthewordsegmentationsystem.Andthenistheclarificationofroughdatatomakeitwellstructured.Eventuallyweexpandtheprimarysentiment1exiconsbasedonwell.processeddata.Andinthisproc
8、ess,wehavecomparedtheadvantagesanddisadv
此文档下载收益归作者所有