预览加载中，请您耐心等待几秒...

基于HowNet的微博文本语义检索研究.docx

预览
1/2
2/2

在线预览结束，喜欢就下载吧，查找使用更方便

下载提示 文本预览

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

基于HowNet的微博文本语义检索研究 1.Introduction Withtheincreasingpopularityofsocialmediaplatforms,microblogs,suchasTwitterandWeibo,havebecomeamajorsourceofinformationforpeople.Millionsofusersusetheseplatformstoexpresstheiropinions,sharenews,andinteractwitheachother.However,themassiveamountofuser-generatedcontentposesaconsiderablechallengeforinformationprocessingandretrieval.Thesemanticunderstandingofmicroblogtexthasbecomeanessentialtaskininformationretrievaltoenableefficientandeffectivecontentanalysis,categorization,searchandrecommendation.Inthispaper,wepresentastudyonusingHowNettoimprovetheperformanceofmicroblogtextsemanticretrieval. 2.RelatedWork Inrecentyears,numerousstudieshavebeenconductedtoaddresstheproblemoftextretrieval.Somestudieshavefocusedontheuseoflexicaldatabasestoimprovetheaccuracyoftextretrieval.Forexample,theWordNetlexicaldatabasehasbeenusedinmanystudiestoenhanceseveralnaturallanguageprocessing(NLP)tasks,suchassemanticsimilaritycomputation,textclassification,sentimentanalysis,andinformationretrieval.TheHowNetlexicaldatabaseisaChinesesemanticknowledgebase,whichhasbeenusedinseveralNLPtasks,suchaswordsensedisambiguation,sentimentanalysis,andopinionmining. 3.HowNetandItsApplications HowNetisalexicaldatabasedevelopedbytheChineseAcademyofSciences.ItisdesignedtoprovideknowledgeabouttheChineselanguageanditsusageincommonlanguage.HowNetincludesmorethan200,000wordswithexplanationsoftheirsemanticandsyntacticpropertiesandrelations.HowNet’sknowledgebasecomprisesconcepts,semanticrelationsbetweenconcepts,lexicalitems,andsyntacticinformation.TheconceptsinHowNetarerepresentedasnodesinahierarchicaltreestructure,wheretherootnodeis‘entity’andtheleavesarespecificentities,suchas‘apple’,‘chair’,‘sky’,andsoon.Theedgesbetweenthenodesrepresentthesemanticrelationsbetweenthem,includingsynonymy,hyponymy,antonymy,entailment,co-hyponymy,andsoon.HowNethasbeenusedinseveralNLPapplications,suchaswordsensedisambiguation,sentimentanalysis,andopinionmining. 4.Methodology Inthisstudy,weusedHowNettocons