利用人工和自動(dòng)生成的資源進(jìn)行中文信息檢索查詢(xún)擴(kuò)展.pdf

1、天津大學(xué)博士學(xué)位論文利用人工和自動(dòng)生成的資源進(jìn)行中文信息檢索查詢(xún)擴(kuò)展姓名：賀宏朝申請(qǐng)學(xué)位級(jí)別：博士專(zhuān)業(yè)：計(jì)算機(jī)應(yīng)用技術(shù)指導(dǎo)教師：何丕廉20020601ABSTRACTTermmismatchbetweenqueriesanddocumentsISafundamentalprobleminChineseInformationRetrieval(取)，whichaffectstheeffectivenessofretrievalresult

2、sInthisthesis，wecarryoutthestudyonqueryexpansionwiththesaurusesgeneratedmanuallyandautomaticallyforChineseIRExperimentalresultsshowthatourmethodsresultsinsignificantimprovementovertheIRwithoutqueryexpansionWeevaluatetheⅡ

3、己systemsands拄atc：giesonthestandardevaluationcollectionsforChineseIR，anddiscussclearlyaboutthebetterindexingunitsofChineseIRFurthermore，wefocusonthestudyofqueryexpansionwiththesaurusesgeneratedmanuallyandautomaticallyforC

4、hineseIRWetrytodoqueryexpansionwiththethesaurusofHowNetdirectlyforChineseIRBecauseoftheinsufficiencyofHowNet，theresultsarenotstatisticallyimprovementWerealizeadecayingCOoccurrencemodel，whichCangeneratethethesaurusofsimil

5、aritybetweenwordsautomaticallyThemodelcangetmoreprecisethesaurusbecauseitconsideraboutthedistancebetweentermswhichiSunlikethecommonCO—oecunencemodelsWeproposeandrealizequeryexpansionbasedonthecontext，whichCanchoosetheexp

6、ansionwordsaccordingtothecontextofthequeryComparedwiththeresultsofIRwithoutqueryexpansion，queryexpansionbasedonthecontextcangetstatisticallysignificantimprovement，whichshowsthatqueryexpansionbasedonthecontextiSasmarterme

7、thodWeproposeandrealizequeryexpansionwitlthethesaurusesgeneratedmanuallyandautomaticallyforChineseIRUsingthecombinationoftheHowNetthesaurusandthe“pseudothesaurus”，queryexpansionbasedonthecontextcangetmoresignificantimpro