中文自動文摘系統(tǒng)的研究與實現(xiàn).pdf

1、南京師范大學(xué)碩士學(xué)位論文中文自動文摘系統(tǒng)的研究與實現(xiàn)姓名：朱荷香申請學(xué)位級別：碩士專業(yè)：計算機應(yīng)用技術(shù)指導(dǎo)教師：曲維光20080401ABSTRACTAutomaticsummarizationisallimportantresearchtopicinthenaturallanguageprocessingInrecentyears，alongwiththevigorousdevelopmentofInternet，theamounto

2、finformationincreasessharplyaswellastheliteraturequantityassumestheexponentialordertogrowAsasupplementarymeanswhichsolvestheoverloadproUlemofinformationautomaticsummarization’svaluegetsmoreandmoresufficientembodiment，itc

3、anhelptoimprovethespeedoftheinformationretrievalandsavethebrowsingtimeofinformationAutomaticsummarizationiscloselyrelativetothesemantics，butthetraditionalstatisticssummarizationextractssentencesthroughtheestablishmentoft

4、hevectorspacemodelbasedonstatisticsofwordfrequencyVectorspacemodel’sbasicassumptionistheirrelevanceamongtheitems，butinrealtextbecauseofthediversityoflanguage，evenifthesameconceptoftenhasmanydifferentformsofexpression，wor

5、dsindivideditemssometimeshavegreatrelevance，nottotallybeindependentInadditionthearticlecontainsalloverallthemegenerallybuttheauthorsometimesillustratethisthemefrommultiplesidesIfonlyextractssummariesinaccordancewiththeim

6、portanceofsentencesinthefullarticle，theresultoRenabstractsthethemewhichhasmoredensitydistributionbutignorestheexistenceofothersubjects，theintegralityisnothighDevotedtothesettlementoftheabovementionedproblems，thispaperado

7、ptsthestatisticalmethodcombiningwiththesemanticknowledge，putsforwardthesolutionbasedonconceptcountingandtextstructurepartitionforautomaticsumrnarizationandrealizestheprototypesystemThespecificworksareasfollows：Firstlythe

8、paperreviewsstudyhistoryofthedomesticandabroadautomaticsummarizationandsummarizescurrentsituationintroducestheoriessuchasvectorspacemode，ChineselexicalanalysisandevaluationofautomaticsummarizationandsoonThen，conceptcount

9、ingtotheresearchofautomaticsummarizationbasedonHITIR—LabTongyiciCilin(Extended)isintroduced，themaximummatchingalgorithmisutilizedtosolvetheproblemofthewordequivocalduringthecourseofconceptobtainingpreliminaryInordertoena

10、blethesummarizationcoverthemaincontentsoftheoriginal，analgorithmfortopicpartitionbasedonacomprehensiveinvestigationofbotlladjacentparagraphicsimilarityandconsecutiveaverageparagraphicsimilarityisputforwardFinallytheautom

11、aticsummarizationsystembasedonconceptcountingandtextstructurepartitionisrealizedIntheevaluationofthesystem，tomakeitmoreobjective，impartialandreasonable，designedevaluatingstandardaimedatevaluationcorpus’scharactersToverif

12、yvalidityandfeasibilityofoursummarizationmethodnecomparedexperimentsonautomaticsummarizationbasedontraditionalmethodconceptcountingmethodandthemethodweputforwardarecarriedoutseparatelyTheexperimentalresultsrevealthat，the

13、methodintextcanreflectthecontentstructureofthearticleeffectivelyandgainbeuereffectthanthetraditionalmethod，especiallywiththeincreaseinthelengthofthesummarization，theeffectofthismethodismoreobviousItisapplicabletothelonga

14、swellastheshortarticlesBesides，thecontrastwithotheravailablesummarizationtoolsshowsthatthemethodweputforwardapproachestheautomaticsummarizationfunctionofIRLab—NLPMLsystemintheHITIR—Labandsurpassesthecorrespondingfunction