中文微博数据净化与情感倾向分析技术的研究与实现

《中文微博数据净化与情感倾向分析技术的研究与实现》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StudyandImplementationonDataCleaningandSentimentAnalysisTechniquesforChineseMicroblogAbstractAsanewinformationcarrier,themicrobloghasbecomepopularandindispensibleinpeople’Sdailylife．Inmicroblog，therearealargenumberofvaluablecommentsoncelebrities，eventsandproducts，whichCanexpressusers’s

2、entimentorientationandplayamoreandmoreimportantroleintheemergenceandpropagationofWebpublicopinion．ForthenewcharacteristicsofChinesemocroblogs，thisthesisstudiesondatacleaning，sentimentorientationanalysis，andtheirrelatedtechniques．Firstly，fortheproblemofspamandnear-duplicatemicroblogs，thethe

3、sisstudiesonthemicroblogdatacleaningapproach．Recently,largenumbersofsparemicroblogsandnear-duplicatemicroblogscovereverycomerofthemicroblogspace．Theyhavebroughtinadverseeffectsontheaccuracyofinformationretrievalandaffectedthecredibilityoffurtheranalysis．Eliminatingthespamandnear—duplicatem

4、icroblogshasbecomeaseriousproblemintherelativeresearcharea．Totacklethisproblem，thecharacteristicsofthespamandnear．duplicatemicroblogsareanalyzedbasedonthestatisticalanalysisresultsofmassivereal．worldmicroblogdata,andafilteringapproachwithfeatureselectionanddoublecontentsimilarit)rdetection

5、formicroblogtextstreamisproposed．TheproposedmethodCanfirstlyfilteroutspammicroblogsthroughtheURLlinks，characterratesandhighfrequencywords．Thenthenear．duplicatemicroblogsareeliminatedthroughthesubsection-basedandindex—basedfilters。Experimentsshowthattheproposedmethodcaneffectivelypurifythem

6、icroblogsbyfilteringoutthespamandnear-duplicatemicroblogs．Secondly，accordingtothecharacteristicsof“straightforwardemotionexpression”，thisthesisstudiesonthesentimentorientationanalysismethodforChinesemicroblog．The“straightforwardemotionexpression'’meansthatpeopleinthemicroblogpostsusedtoemp

7、loyemoticons，interjections，anddegreeadverbstoexpression．TheexistingsentimentlexiconsandprevioussentimentanalysismethodsCanbeappliedtosentimentanalysistaskforChinesemicroblogs，butthesemethodsusuallyignorethenewcharacteristicsofthemicroblogcontent．Thereisalackof