搜索
摘要随着计算机的广泛应用和互联网技术的迅猛发展，众多的信息检索系统被开发出来，方便用户获取其感兴趣的内容。数字化的信息资源一方面为人们的工作和生活带来了帮助，另一方面，大量的信息又使人们迷失在信息的海洋中。造成这种结果的原因在于目前的信息检索系统主要是基于关键词匹配。如何组织和提供信息就成为信息检索系统要解决的关键问题。本体因为所具有的良好概念层次结构和对逻辑推理的支持，在信息检索，特别是在基于知识的检索中得到了广泛地应用，成为研究热点。本文以本体为基础，结合传统信息检索技术构建了一个基于本体的知识检索框架OKRF。首先，本文归纳了传统信息检索的不足，分析了基于语义的信息检索的特点以及本体在其中发挥的作用，总结了语义检索领域的国内外研究现状。其次，本文研究了语义网、本体的建模原语及层次结构和信息检索领域的相关技术，提出了基于本体的知识检索框架OKRF，介绍了OKRF框架的两大系统：知识库构建系统和查询系统。然后，阐述了OKRF框架的系统构成，研究了框架中主要模块所使用的相关技术。在知识库的构建部分，提出了本体构建的V-模型，介绍了基于本体词汇表的类型标注，并给出了关系三元组的抽取方法。在查询系统部分，提出了两种查询扩展的方法，并给出了查询结果排序的计算公式。最后，本文将OKRF框架应用在石油测井领域，基于OKRF框架设计并实现了一个测井知识管理系统WLKMS。关键词：知识检索，语义检索，语义网，本体，框架基于本体的知识检索框架的研究IIABSTRACTWiththeextensiveuseofcomputerandInternettechnology,alargenumberofinformationretrievalsystemsweredevelopedtoprovideuserswithinterestingcontents.Ononehand,digitalinformationresourcesareconvenientforpeople,ontheotherhand,somuchinformationretrievedmakepeopleconfused.Thereasonforconfusionmentionedaboveisthatthetechniqueusedbyinformationretrievalsystemisstillkeywordsmatching.Sohowaninformationretrievalsystemorganizesandprovidesinformationbecomesakeyissue.Atpresent,becauseofontology’sgoodconceptofhierarchyandgoodsupportuponthelogicalreasoning,itbecomesmoreandmorewidelyusedininformationretrievalfield,especiallyinknowledgeretrievalfield.ThisdissertationproposesanOntology-basedKnowledgeRetrievalFramework(OKRF)bycombiningtheontologytheorywiththetraditionalsearchtechnology.Firstly,thisdissertationsummarizestheshortcomingsoftraditionalinformationretrieval,analyzesthecharacteristicsofsemanticretrievalandtherolethatontologyplayedin,summarizesthehomeandabroadresearchonsemanticfield.Secondly,thisdissertationstudiestheSemanticWeb,ontologymodelinglanguageandhierarchicalstructure,informationretrieval.ThenthisdissertationproposesthestructureoftheOKRF,introducestwomajorsystemsoftheframework:knowledgebaseconstructionsystemandinquirysystem.Thirdly,thisdissertationexplainsthestructureofOKRFandstudiestherelatedtechniquesusedbythemainmoduleoftheframework.Inthepartofknowledgebaseconstructionsystem,thisdissertationproposestheV-modelforbuildingontology,introducestypemarkingbasedonontologyvocabularyandexplainstheextractionmethodofthetri-tuplerelationship.Inthepartofinquirysystem,thisdissertationproposestwomethodsofqueryexpansionandtheformulaforrankingtheresults.Finally,thisdissertationrepresentsthedesignandimplementationofawelllongingKnowledgeManagementSystems(WLKMS)whichisbasedonOKRF.Keywords:knowledgeretrieval,semanticretrieval,semanticweb,ontology,framework南京航空航天大学硕士学位论文V图表清单图1.1论文组织结构·························································································································4图2.1语义网体系结构·····················································································································5图2.2本体信息的层次模型·············································································································7图2.3OKRF框架的系统结构图······································································································10图3.1知识库构建的基本流程········································································································13图3.2本体构建的V-模型概念图····································································································15图3.3本体构建的具体V-模型········································································································15图3.4本体构建的迭代过程············································································································18图3.5类型标注的基本流程············································································································20图3.6解析结果依存树····················································································································23图4.1前缀树结构的字典组织形式·································································································27图4.2后缀树结构的字典组织形式·································································································28图4.3词语、概念和对象三者之间的关系·····················································································30图4.4CBQE算法流程·····················································································································31图4.5RCBQE算法流程···················································································································32图4.6词语、概念和文档之间的关系·····························································································34图4.7一个概念下属于相同数目文档的两个词语举例··································································35图5.1WLKMS的系统结构图··········································································································39图5.2知识库核心概念层次图········································································································41图5.3Protégé的界面截图················································································································42图5.4本体主要类关系图····························································