資源描述:
《語義網(wǎng)中本體的構(gòu)建技術(shù)》由會員上傳分享,免費(fèi)在線閱讀,更多相關(guān)內(nèi)容在學(xué)術(shù)論文-天天文庫。
1、摘要在語義網(wǎng)(Semanticw曲)研究中,本體(Ontology)占有極其重要的地位,作為一種有效表現(xiàn)概念層次結(jié)構(gòu)和語義的模型,本體被廣泛地應(yīng)用到計(jì)算機(jī)科學(xué)的眾多領(lǐng)域。本體構(gòu)建包括兩個核心問題:一是如何支持本體構(gòu)建,二是如何確保本體正確構(gòu)建。針對問題一,本文研究開發(fā)了本體構(gòu)建系統(tǒng)OntoMAN,以人工的方式支持本體構(gòu)建;探討基于知識元數(shù)據(jù)驅(qū)動本體自動構(gòu)建。對于問題二,本文提出了基于判別規(guī)則識別本體中不滿足概念的最小不一致知識子集(MinimalUnsatisfiability-preservingSub.TBoxes,MUPS)以確保本體正確構(gòu)建。構(gòu)建工具O
2、ntoMAN在實(shí)現(xiàn)基本的本體實(shí)體管理的基礎(chǔ)之上,集成不滿足概念分析機(jī)制,協(xié)助本體建模人員構(gòu)建本體。將聚類算法與ODP(OpenDirectoryProject)目錄有機(jī)結(jié)合,給出了一種基于知識資源元數(shù)據(jù)的本體自動構(gòu)建方法。根據(jù)元數(shù)據(jù)對文檔進(jìn)行聚類形成本體概念,將生成的概念分別映射到ODP中確定概念間的層次關(guān)系,生成初始本體;根據(jù)內(nèi)聚性和相關(guān)性的變化進(jìn)行自適應(yīng)本體學(xué)習(xí),實(shí)現(xiàn)本體更新和概念豐富。實(shí)驗(yàn)結(jié)果表明了方法的有效性。識別不滿足概念C的最小不一致知識子集是修復(fù)該不滿足概念的前提??紤]到不滿足概念的傳遞性,將MUPS分為三類:完全依賴于C(MUPsf)、傳遞依賴
3、于C(MuPst)和不確定依賴于C(MuPsu)。分析上述分類結(jié)果,本體建模人員和領(lǐng)域?qū)<铱梢悦鞔_下一步修復(fù)工作的方向和步驟。提出基于兩個對等轉(zhuǎn)換,即公理細(xì)化和本體約簡,并應(yīng)用三個判別規(guī)則,識別不滿足概念c的MUPS的方法。實(shí)驗(yàn)結(jié)果表明,所得分類結(jié)果對于從修復(fù)角度評價本體質(zhì)量以及指導(dǎo)修復(fù)工作都具有重要意義。關(guān)鍵字:語義網(wǎng)本體構(gòu)建最小不一致知識子集IAbstract111SemanticWebresearch,ontologyisextremelyimportant.Asapromisingmodelforpresentinghierarchyandsemant
4、icmeaningofconcepts,itiswidelyconcernedandextensivelyappliedtomanyfieldsincomputerscienceandtechnology.Thecoreissuesofontologyconstructionincludetwopoints:First,howtosupportontologyconstruction,thesecondiShowtoensurethattheontologyconstructediscorrect.Forthefirstissue,theontologycon
5、structiontoolOntoMANisdevelopedtosupportbuildingontologymanually,andontologyautomaticallybuildingisdiscussedbasedonmetadataofknowledgeresources.Forthesecondone,theapproachforidentifyingMinimalUnsatisfiability-preservingSub.TBoxes(MUPS)basedondiscriminationrulesiSproposedtoensurecorr
6、ectconstruction.TheontologyconstructiontoolOntoMANnotonlyachievesthebasicmanagementofontologyentity,butalsointegratesthemechanismofanalyzingtheunsatisfiableclass,whichhelptobuildontology.Anautomaticontologyconstructionapproachbasedonmetadataofknowledgeresourcesisproposedbyintegratin
7、gclusteringalgorithmandOpenDirectoryProject(ODP).Ontologyconceptsaregeneratedbyclusteringdocumentsbasedontheirmetadata,andconcepthierarchyisformedbasedonthehierarchyofmappedconceptsinODEAdaptiveontologylearningisconductedtoupdatetheontologyandenrichontologyconceptsbasedonthechangeso
8、fcohesionandcorrela