受限語境下文本語義相似度計算研究及應(yīng)用.pdf

1、東南大學(xué)碩士學(xué)位論文受限語境下文本語義相似度計算研究及應(yīng)用姓名：潘國清申請學(xué)位級別：碩士專業(yè)：計算機應(yīng)用技術(shù)指導(dǎo)教師：滕至陽20070501ABSTRACTExaminationsystemisallimportantareaincunelIteducationsystemandvariousevaluationandselectioninourcountryWfittenexaminationcallbcdividedintoobje

2、ctiveandsubjectiveq增sIio惦ObjectivequestionscanbescoredbycomputerwhilesubjectiveonesstillcannotbeItisessentialtesearchthemethodstoscorethesubjectivequestionsbycomputertosavetheworkloadoftheteachersThekeytodevelopingsuchsy

3、stemis幻findthemethodtocalculatethetextssemanticsimilarityTheresearchisconcernedwiththeChineseversionsemanticsimilaritycalculationmethodinthelimitedcontext，ItputsforwardakindofmethedtoexpressChinesetextsemanticLanguageFra

4、gmentThefollowingarethemainpoints(1)111eWaditionalvectorspacemodel(VSM)semanticrepresentationhasdeficiencies吲ngkeywordsfeatttreitemsnearticleputsforwardtheLanguageFragmentmethodonthebasisoftheconceptofchunLThestructureof

5、LanguageFragmentis”keywordssubsidiary”ItexpoundsandprovestheapplicationofLanguageFragmentbasedonthefeasibilit／ofthetheoryandthenecessityofthetechniqueItComparestheLanguageFragmentwithkeywordsandtheconceptofchunkpointsout

6、thedifferencesamongthemandshowsthefeaturesandadvantagesoftheLanguageFragmentItwastheLanguageFragmenttoexpresstextfeaturesandtextsemanticsItestablishestextsemanticsusingVSM，theweightsisabsolutetermsFrequency，usingvectorco

7、sinearIgletocalculatesemanticsimilaritiesbetweentexts(2)Basedonthetextcorpusof”operatingsystem”bycalculatingthemutualinformationbetweentwowordsselectwordswhichmeettheselectedthresholdcondili∞gleedsofthresholdvalue，usingt

8、hedualgrammaticalrules，establishcorrelationdatabasebetweentwowordswithinthelimitedcontexttodeterminethesubsidiarywords(3)EstablishaChinesewordpusmarkercollectionincluding40Symbolsaformwordposmarkercollectionincludingl1sy

9、mbolsemptyWOrdSastopwordlistscontains649wordsandaChinesedictionaryofsynonymscontalm31，621wordswhichareusedtohandleChinesetextprocessingFortherespectivesubterms，partofspeechtagging∞ctOlSsuchasremovedofhighfrequencystopwor

10、dsandsynonyms11忙systemofautomat／callyscoringsubjectivequestionsaboutthecourseof”operatingsystem”hasbecaestablishedTheexperimentinwhich1036textpapersscoredbythesystemofautomaticallyscoringcontrastingthesescoredmanuallysho

11、wedthatthecorrectrateoftheformeronewas8524％Tbecomparativeexperimentwiththesimilarsystemofthenation15keyscientificandtechnologicalprojectsprovedthattheeffectivesofthissystemisequaltothecontrastsystemKEYWORDS：VSMfeatureite