基于ICVSM的摘要抽取算法研究.pdf

1、河南科技大學(xué)碩士學(xué)位論文基于ICVSM的摘要抽取算法研究姓名：郭志兵申請(qǐng)學(xué)位級(jí)別：碩士專業(yè)：計(jì)算機(jī)應(yīng)用技術(shù)指導(dǎo)教師：黃廣君20091201摘要IISubject:ResearchonSummarizationAbstractAlgithmbasedonImprovedCVSMSpecialty:TheTechnologyofComputerApplicationName:GuoZhiBingSupervis:HuangGuangJunv

2、iceprofessABSTRACTSummarizationabstractisatechniqueofInfmationAbstractwhichisaresponsetothemoderninfmationwld.Itcanextractsentencesfromalargetextquicklyaccuratelywhichcanexpressthemeaningofthetexttogeneratesummarizationh

3、elppeopletogainusefulinfmationefficiency.FirstlythethesisintroducestheresearchactualityrelatedtechniquethenaccdingtothedisadvantageoftheChinesesummarizationabstractalgithmwhichbasedonthecombinedofstatisticsemanticpropose

4、daimprovedabstractionalgithm.Thenewalgithmimprovedthepreviousmethodfromthefollowingtwoaspects.AccdingtotheambiguityofChinesewdsthisthesisputfwardanimprovedalgithmofwdsensedisambiguation.InthismethodweuseHowcpusbuildsthep

5、rimitivecooccurrencefrequencydatabaseasthebasisfwdsensedisambiguation.Whencalculatethecrelationcoefficientofwdsensecontextconsiderthecrespondingrelationshipofthefourkindsofprimitiveshavedifferenceabilityonsemanticexpress

6、ingconsidertwodistancefactswhichaffecttheexpressionofsemantictheoneisthespacedistancebetweentheacterwdsthemaltvocalwdtheotheristhespacedistancebetweenthecurrentlymaltyvocalwdthesamemaltvocalwdwhichhasbeenedsenseatthelate

7、st.AccdingtotheindependencybetweenitemsoftheCVSMthisthesisputfwardavagueconceptequivalenceclasspartitioningalgithmwhichbasedonclusteringconcept.Consideringtheactualitysignificancethisalgithmcombinedtheconceptswhichhaveno

8、distinctdifferenceinsemanticsexpressinghavegreatsimilarity.UsethemusterofconceptualastheCVSM’sitemsinsteadofthesingleconceptualusetheimprovedCVSMtoexpressthetextthentranslatethetextintodatamecrectsothattogenerateameconci