欢迎来到天天文库
浏览记录
ID:36576210
大小:3.14 MB
页数:61页
时间:2019-05-12
《本体知识库的自然语言查询接口研究》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、上海交通大学硕士学位论文本体知识库的自然语言查询接口研究姓名:熊淼申请学位级别:硕士专业:计算机应用技术指导教师:俞勇20070101上海交通大学硕士学位论文自然语言查询里面的词语与本体实体之间建立映射,然后使用本文提出的轻量级WSD(WordSenseDisambiguation,单词歧义消除)算法结合整个查询语句的上下文来确定产生的本体三元组。最后,SPARQL查询组装模块将本体三元组和修饰符抽取模块的结果生成可以执行的SPARQL查询语句。通过对该研究领域上的经典数据集上的实验表明,本文提出的方法在转化的查询的数量和质量都较以往系统有明显提高。同时,这些实验也发现了本文的方法在处理复杂的
2、自然语言查询方面存在的某些不足(例如对推理的支持)。这些不足是本文下一步工作的重点。关键词语义网,本体,自然语言查询接口,SPARQL,形式化查询5上海交通大学硕士学位论文STUDYONNATURALLANGUAGEINTERFACETOONTOLOGYKNOWLEDGEBASESABSTRACTWiththedevelopmentofsemanticwebtechnologies,moreandmoreontologyknowledgebases(OKB)areconstructed.However,inordertogainknowledgefromOKBs,formallogicquer
3、iesarenecessary.OrdinaryusersrarelyusethesimplestBooleanquery,notmentionthecomplexlogicexpression.Therefore,buildinganaturallanguageinterfacetobridgethegapbetweenordinaryusersandformalizedOKBscannotonlyenableuserstoutilizeexactformalknowledgeasknowledgesharing,butalsoenlargetheeffectsofOKBsandconseq
4、uentlyencouragepeopletocreatemoreOKBs.InordertohelpusersexpresstheirinformationneedsvianaturallanguagethattheyarefamiliarwithandgaininformationfromOKBs,thispaperproposesaframeworkandsystembasedontheexistingworkofsemanticwebtechnologies.Thispaperintroducesanovelmethodviacombinationofdeepandshallowsyn
5、taxanalysismethods6上海交通大学硕士学位论文basedonthesurveyofrelatedworks.WiththehelpofWordNetandothertools,alexiconisbuiltautomaticallywhenanOKBischosen.ThelexiconbridgesthewordgapbetweenusersandOKBsbyprovidingmappingbetweenusers’wordsandtheentitiesinOKB.Thedeepsyntaxprocessorfirstlytranslatessyntaxtreegenerat
6、edbyNLP(NaturalLanguageProcessing)parsersintoquerytriples.Andthentranslatesthemintoontologytriplesthroughtriplemappingalgorithm.Atthemeantime,theshallowsyntaxprocessorfirstlyacquiresallpossiblemappingsandthenresolvesthemappingproblemwithanovellightweightWSD(WordSenseDisambiguation)algorithm.Atlast,t
7、ogetherwiththeresultsproducedbymodifierextractor,theSPARQLgeneratorcreatesfinaloutputSPARQLqueries.Withtheexperimentsontheclassicaldatasetsofnaturallanguageinterfaceresearchdomain,thetransformationper
此文档下载收益归作者所有