欢迎来到天天文库
浏览记录
ID:28054480
大小:92.00 KB
页数:11页
时间:2018-12-07
《基于本体和局部查询反馈的微博查询扩展算法》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、基于本体和局部查询反馈的微博查询扩展算法M杜军平赖金财梁美玉王巍罗盎北京邮电大学计算机学院智能通信软件与多媒体北京市重点实验室新浪网技术(中国)有限公司摘要:传统的基于关键词匹配的查询方法因查询词短少,微博博文短小,容易引起歧义性,对查询效率有较大影响.提出一种基于木体和局部查询反馈的微博查询扩展算法,首先结合安全领域文档构建安全领域本体知识库,然后利用本体提供的语义知识对初始查询词进行扩展,再结合局部查询反馈对候选扩展词集进行筛选,最后通过二次奔询和迭代操作得到最终奔询结果.实验结果表明,基于本体和局部查询反馈的微博查询扩展算法比
2、基于关键词的查询扩展算法、基于本体的查询扩展算法和基于“伪相关反馈”的查询扩展算法有更好的查全率和查准象关键词:本体;微博;共现分析;查询扩展;作者简介:杜军平,E-mail:junpingdu@126.com收稿日期:2017-09-15基金:国家自然科学基金(61532006,61320106006,61502042)MicroblogqueryexpansionalgorithmbasedonontologyandlocalqueryfeedbackGongHaoDuJunpingLaiJincaiLiangMeiyuWang
3、WeiLuoAngBeijingKeyLaboratoryofIntelligentTelecommunicationSoftwareandMultimedia,SchoolofComputerScience,BeijingUniversityofPostsandTelecommunications;SINACorporation;Abstract:Thepurposeofthisworkistomeasuretheefficiencyofinformationretrieval(IR)inmicroblogbyusingquery
4、expansionbasedonontologyandlocalqueryfeedback.Firstly,ontologyknowledgebaseofsecuritydomainiscreatedbysecuritydomaindocuments.Then,theontologyisexpendedbyusingthesecuritydomainterminologyextractedfrommicroblogdocuments.Thus,theexpendedontologyconsiststwobroadcategories
5、,sixsubclassesandmorethanfiftyconcepts.Secondly,thequerywordisexpendedbythesemanticknowledgeprovidedbytheexpandedontology.AndtheLucencesearchengineisusedforinitialretrieval.Bycalculatingmicroblogheatandtimecorrelation,localmicroblogdocumentsaregottofiltertheexpansionwo
6、rds.Finally,combiningtheweightofeachcandidateexpansionwordinontologyqueryexpansionandlocalqueryfeedbackco-occurrenceanalysis,thefilterfunctioniscreatedtoselectthefinalexpansionwords-Thefinalresultsarcgotbyiterativeoperationandsecondaryretrieval.Inordertochecktheaccurac
7、yofthemicroblogqueryexpansionalgorithmbasedonontologyandlocalqueryfeedback(OFQE),keywordsqueryexpansionalgorithm(KQE),ontologyqueryexpansionalgorithm(OQE)andpseudorelevancefeedbackqueryexpansionalgorithm(PRPQE)areusedtocomparetheefficiencyofmicrobloginformationretrieva
8、l.Multiplequerywordsandtheircombinationsareusedforretrieval.TheexperimentalresultsaretheaveragescoresoftopNresultsbym
此文档下载收益归作者所有