資源描述:
《基于語義處理技術的信息檢索模型研究》由會員上傳分享,免費在線閱讀,更多相關內(nèi)容在學術論文-天天文庫。
1、浙江大學計算機學院博士學位論文基于語義處理技術的信息檢索模型研究姓名:王瑞琴申請學位級別:博士專業(yè):計算機科學與技術指導教師:孔繁勝20090401浙江人學博士學位論文摘要優(yōu)先返回與查詢語義相關性強的文檔供用戶瀏覽。4.本文對如何滿足不同用戶的個性化查詢需求進行了研究,提出了一種語義加強的個性化信息推薦方法。該方法綜合利用語義數(shù)據(jù)源和歷史評分數(shù)據(jù)進行混合推薦,語義數(shù)據(jù)源的引入解決了傳統(tǒng)協(xié)同過濾系統(tǒng)的數(shù)據(jù)稀疏性和冷啟動問題。另外,為了提高推薦系統(tǒng)的可擴展性和實時性,在數(shù)據(jù)的離線預處理階段,本文使用數(shù)據(jù)挖掘方法對用戶和項目進行了模糊聚類。關鍵字:信息檢
2、索,語義關聯(lián),隱式反饋,詞義消歧,查詢擴展,語義相關性,查詢優(yōu)化,聚類,個性化推薦浙江大學搏上學位論文AbstractWeareinaninformationagethatmainlycharacterizedbyinformationexplosion,andinformationretrievaltechniquesarenowchallengedalotbymorefrequentInteractinformationupdating,aswellasincreasinguserdemandformoreprecisesearchresult
3、s.Semanticsearchtechnique,fortunately,isahopefulwaythatleadstothekeytotheissueoffindingexactinformationfrommassnumberofthemeffectively.However,asaresultoftheincompleterealizationofsemanticwebtechnique,recentstudyhasbeenmorefocusedonsemanticretrievaltechniqueintransitionperiod,
4、makingitahottopicofresearch.SeveralkeyproblemsinInformationRetrieval(IR)domainareaddressedandanovelSemanticProcessingTechnologybasedInformationRetrieval(SPTIR)modelisproposedinthisdissertation.SPTIRisanextensiononQueryExpansion(QE)andSearchResultre-Ranking,whichconsistsoffourp
5、arts,namelysemanticqueryexpansionbasedonWordSenseDisambiguation(WSD),queryoptimizationbasedonwordsemanticrelatedness,searchresultsre—rankingbasedondocumentsemanticrelevance,andsemanticenhancedpersonalizedinformationrecommendation.Firstly,inthecontextofkeyword-basedsearchengine
6、,awell—structuredandgood-meaningfuluserquerynotonlyexpressesL1ser’spersonalneedsprecisely,butalsoguaranteestheQS(QualityofService)requirementforinformationretrieval.Startingwiththeissueofsemanticassociationsofquerykeywords,supplementedbyimplicitfeedbacktechnique,andusingunsupe
7、rvisedWordSenseDisambiguation,thisdissertationpresentsatechniquethatmapsquerykeywordstoontologyconcepts,andasemanticqueryexpansiontechniquebasedonconcept—wordassociation.TheWSDbasedsemanticqueryexpansionsolvestheproblemofnotwellunderstandinguser’Squeryintensionintraditionalret
8、rievalsystems.Secondly,forthosequerykeywordsthatfailtodisambi