資源描述:
《基于中文微博的情感-分析研究》由會員上傳分享,免費在線閱讀,更多相關(guān)內(nèi)容在教育資源-天天文庫。
1、華中科技大學碩士學位論文AbstractMicroblogisbecomingamostpopularinternetapplication.Accordingtothestatis-tics,morethan100milliontweetspublichedineveryday.Thesetweetsnotonlyconveythedescriptionoffacts,butalsocontaintheemotionalstatesofmassivemicroblogusers.Andtheseemotionalinformationsmaybehelpfo
2、rusertodecidewhetherbuyaproduct,provideveryimportantreferencevalueforcompaniestomakemarketstrategy,andevenmakemassivedataavailableforgovernmenttomonitoringpublicopinion.Inlightofthis,weproposedasentimentanalysismethodbasedonacombinationofsyntacticdependenciesandtextclassificationtec
3、hniquesforChinesetweets.Themethodadoptsthesyntacticdependenciestoperformsentimentanalysis,atthesametime,com-putesaconfidenceforeverytweet.Choosentweetswhichconfidenceaboveacertainthresholdastrainingsamples,trainatwo-stepsentimentclassifierbyusingthecontentfeaturesandmediafeaturesoft
4、weets.Finally,classifythesentimentorientationoftweetsagain.Inaddiation,wealsoproposedamethodthatservescommonemoticonsasthesen-timentclasslabelsoftweetsandimplementsanincrementallearningmethodtotackletheproblemofreal-timesentimentanalysis.Experimentalresultsshowthattheproposedmethodd
5、ramaticallyimprovesthepre-cisionandtherecallby6%and3%repectivelycomparedtothemethodthatonlybasedonsyntacticdependencies.Andtheperformanceofourtwofeaturesetsarealsobetterthanunigramfeatures,theprecisionandtherecallbothare88%intermofsubjectiveclassifier,andtheyare72.1%and71.5%forsenti
6、mentclassifier.Apartfromthis,themediafeaturesaregoodfortracklingtheproblemofreal-timesentimentanalysis.Keywords:ChineseMicroblog,SentimentAnalysis,Syntacticdependencies,TextClassi-ficationII華中科技大學碩士學位論文目錄摘要·············································································
7、······IABSTRACT···········································································II1緒論1.1課題研究背景···································································(1)1.2課題的研究目的和意義·······················································(2)1.3國內(nèi)外研究現(xiàn)狀····························
8、···················