Explaining the Stars Weighted Multiple-Instance Learning for Aspect-Based Sentiment Analysis

Explaining the Stars Weighted Multiple-Instance Learning for Aspect-Based Sentiment Analysis

ID:40715595

大小:861.94 KB

页数:12页

时间:2019-08-06

Explaining the Stars Weighted Multiple-Instance Learning for Aspect-Based Sentiment Analysis _第1页
Explaining the Stars Weighted Multiple-Instance Learning for Aspect-Based Sentiment Analysis _第2页
Explaining the Stars Weighted Multiple-Instance Learning for Aspect-Based Sentiment Analysis _第3页
Explaining the Stars Weighted Multiple-Instance Learning for Aspect-Based Sentiment Analysis _第4页
Explaining the Stars Weighted Multiple-Instance Learning for Aspect-Based Sentiment Analysis _第5页
资源描述:

《Explaining the Stars Weighted Multiple-Instance Learning for Aspect-Based Sentiment Analysis 》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、ExplainingtheStars:WeightedMultiple-InstanceLearningforAspect-BasedSentimentAnalysisNikolaosPappasAndreiPopescu-BelisEPFLandIdiapResearchInstituteIdiapResearchInstituteRueMarconi19RueMarconi19CH-1920Martigny,SwitzerlandCH-1920Martigny,Switzerlandnikolaos.pappas@idiap.chandrei.pop

2、escu-belis@idiap.chAbstractandthemusic,butnottoomuchtheactors.De-terminingtheratingsofeachaspectautomaticallyThispaperintroducesamodelofmultiple-isachallengingtask,whichmayseemtorequireinstancelearningappliedtothepredic-theengineeringofalargenumberoffeaturesde-tionofaspectratings

3、orjudgmentsofsignedtocaptureeachaspect.Ourgoalistoputspecificpropertiesofanitemfromuser-forwardanewfeature-agnosticsolutionforana-contributedtextssuchasproductreviews.lyzingaspect-relatedratingsexpressedinatext,Eachvariable-lengthtextisrepresentedbythusaimingforafiner-grained,deepe

4、ranalysisofseveralindependentfeaturevectors;onetextmeaningthanoverallsentimentanalysis.wordvectorpersentenceorparagraph.Currentstate-of-the-artapproachestosentimentForlearningfromtextswithknownas-analysisandaspect-basedsentimentanalysis,at-pectratings,themodelperformsmultiple-tem

5、pttogobeyondword-levelfeatureseitherbyinstanceregression(MIR)andassignsim-usinghigher-levellinguisticfeaturessuchasPOSportanceweightstoeachofthesentencestagging,parsing,andknowledgeinfusion,orbyorparagraphsofatext,uncoveringtheirlearningfeaturesthatcapturesyntacticandseman-contri

6、butiontotheaspectratings.Next,ticdependenciesbetweenwords.Onceanappro-themodelisusedtopredictaspectratingspriatefeaturespaceisfound,theratingsaretypi-inpreviouslyunseentexts,demonstratingcallymodeledusingalinearmodel,suchasSup-interpretabilityandexplanatorypowerforportVectorRegre

7、ssion(SVR)with`2normforitspredictions.WeevaluatethemodelonregularizationorLassoRegressionwith`1norm.sevenmulti-aspectsentimentanalysisdataBytreatingatextglobally,thesemodelsignorethesets,improvingoverfourMIRbaselinesfactthatthesentencesofatexthavediversecon-andtwostrongbag-of-wor

8、dslinearmod-tributionstotheoverallsentim

当前文档最多预览五页,下载文档查看全文

此文档下载收益归作者所有

当前文档最多预览五页,下载文档查看全文
温馨提示:
1. 部分包含数学公式或PPT动画的文件,查看预览时可能会显示错乱或异常,文件下载后无此问题,请放心下载。
2. 本文档由用户上传,版权归属用户,天天文库负责整理代发布。如果您对本文档版权有争议请及时联系客服。
3. 下载前请仔细阅读文档内容,确认文档内容符合您的需求后进行下载,若出现内容与标题不符可向本站投诉处理。
4. 下载文档时可能由于网络波动等原因无法下载或下载错误,付费完成后未能成功下载的用户请联系客服处理。