中文知识图谱构建的关键技术.pdf

中文知识图谱构建的关键技术.pdf

ID:20839808

大小:3.23 MB

页数:28页

时间:2018-10-17

中文知识图谱构建的关键技术.pdf_第1页
中文知识图谱构建的关键技术.pdf_第2页
中文知识图谱构建的关键技术.pdf_第3页
中文知识图谱构建的关键技术.pdf_第4页
中文知识图谱构建的关键技术.pdf_第5页
资源描述:

《中文知识图谱构建的关键技术.pdf》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、IntelligentSearchEngineandRecommenderSystemsbasedonKnowledgeGraph阳德青复旦大学知识工场实验室yangdeqing@fudan.edu.cn2017-07-13Background•KnowledgeGraphexhibitsitsexcellentperformancethroughtheintelligentapplicationsbuiltonit•AstypicalAIsystems,Searchengineandrecommendersystemareverypopularandpromisin

2、gintheeraoflargedata•ManypreviousliteraturesandsystemshaveprovedKG’smeritsonsuchAI’sapplicationsKG-basedSearchEngineTheHistoryofSearchEngine•1st:category-based•Thekeywordofhighclick•Yahoo,hao123frequencyarerankedhigher•2nd:IR-based•Thepagescontainingthekeywordsofmoreweights•Keyword-base

3、d,vectorresultinarerankedhigherspace,Booleanmodel•Thepageshavingmore•3rd:link-basedimportantin-linksareranked•PageRank(Google)higherHowever,howtohandleitifuserswanttosearchsomethingnewortheonesoflongtail?TheHistoryofSearchEngine•WhenKnowledgeGraphemergesthing/entitykeyword/stringrelevan

4、tentitiesTheHistoryofSearchEngine•Search4.0:user-basedsearchengine•Reallyunderstandthesearchintentofusers,especiallyforthoserareentities•Directlyreturnpureanswers(entities)insteadofrelevantwebpages•Provideplentiful/relevantresultswithhighconfidence/interpretabilityintelligentsearchengin

5、eItisactuallyarecommendation.KnowledgegraphcanhelptheengineachievethesegoalsEntitySuggestionwithConceptualExplanation[1]•MorethansearchiPhone6PlusorCananysearchenginereturnsuchresults?•Problemstatement•Givenaqueryqdescribedbyasetofentities,theengineshouldreturnsomerelevantentitiesalongw

6、ithsomefine-grainedconceptswhichcanwellexplaintheresultsEntitySuggestionwithConceptualExplanation•Whataretherelevantentitiesandfine-grainedconcepts?GrowingMarketEmergingeconomyBRICCountryWhatentitiesshouldbesuggested?CompanyChinesecompanyChineseinternetgiantEntitySuggestionwithConceptua

7、lExplanation•PreviousWorks•1.Co-occurrence(textorquerylog),e.g.GoogleSet,BayesianSet•2.Weblists,e.g.SEAL,SEISA•3.Overlapofproperties•Weakness•1.Co-occurrence≠conceptualcoherence•2.Membershipoflists→toogeneral/specificconceptsorarandomlycombinedlist•3.Overlapofproperties≠concept

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。