资源描述：
《Sentiment Analysis on the People’s Daily 》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。
1、SentimentAnalysisonthePeople’sDailyJiweiLi1andEduardHovy21ComputerScienceDepartment,StanfordUniversity,Stanford,CA94305,USA2LanguageTechnologyInstitute,CarnegieMellonUniversity,PA15213,USAjiweil@stanford.eduehovy@andrew.cmu.eduAbstractAutomaticopinionextractionfromnewspa
2、perssuchaspeople’sdailycanfacilitatesociologistsWeproposeasemi-supervisedbootstrap-’orpoliticalscientists’researchorhelppoliticalpingalgorithmforanalyzingChina’sfor-punditsintheirdecisionmakingprocess.WhileeignrelationsfromthePeople’sDaily.ourapproachappliestoanynewspape
3、rinprinci-Ourapproachaddressessentimenttar-ple,wefocushereonthePeople’sDaily2(Renmingetclustering,subjectivelexiconsextrac-Ribao),adailyofficialnewspaperinthePeople’stionandsentimentpredictioninaunifiedRepublicofChina.framework.Differentfromexistingalgo-Whilemassivenumbero
4、fworkshavebeenin-rithmsintheliterature,timeinformationtroducedinsentimentanalysisoropiniontargetisconsideredinouralgorithmthroughaextractionliterature(fordetails,seeSection2),ahierarchicalbayesianmodeltoguidethefewchallengeslimitpreviouseffortsinthisspe-bootstrappingappr
5、oach.Wearehopefulcifictask:First,theheavyuseoflinguisticphe-thatourapproachcanfacilitatequantita-nomenoninthePeople’sDailyincludingrhetoric,tivepoliticalanalysisconductedbysocialmetaphor,proverb,orevennicknames,makesex-scientistsandpoliticians.istingapproacheslesseffectiv
6、eforsentimentin-ferenceasidentifyingtheseexpressionsisahard1IntroductionNLPprobleminnature.“Wehavenopermanentallies,nopermanentfriends,butonlySecond,aswearemoreinterestedinthedegreepermanentinterests.”ofsentimentratherthanbinaryclassification(i.e.,-LordPalmerstonpositivev
7、ersusnegative)towardsanentity(e.g.Newspapers,especiallythoseownedbyofficialcountryorindividual)inthenewsarticle,straight-governments,e.g.,PravdafromSovietUnion,forwardalgorithmstoapplywouldbedocument-orPeople’sDailyfromP.R.China,usuallypro-levelsentimentanalysisapproaches
8、suchasvectorvidedirectinformationaboutpoliciesandview-machine/regression(Pangetal.,2002)orsuper-pointso
