基于Co-training方法的中文組塊識別的研究.pdf

1、東北大學碩士學位論文基于Cotraining方法的中文組塊識別的研究姓名：劉世岳申請學位級別：碩士專業(yè)：計算機系統(tǒng)結(jié)構(gòu)指導(dǎo)教師：張俐20041201東北大學碩士學位論丈ResearchonChineseTextChunkingBasedonCotrainingAbstractSyntaxanalysisanalysisalsocalledisalwaysabasictaskinthenaturallanguageprocessingsh

2、allowparseorchunkidentificationbecomesahotspotinpartthenaturallanguageprocessing.Nowchunkidentificationiswidelyusedinmanyfieldsofnaturallanguageprocessingespeciallyintheexamplebasedmachinetranslation(EBMT)inwhichchunkide

3、ntificationisoneofmajortechniquesWiththedevelopmentofthetheoryofmachinelearning，machinelearningmethodesbecomesmoreandmoreattractiveinthenaturelanguageprocessingespeciallyunsupervisedandsemisupervisedmachinelearningmethod

4、s.ItliesontwopointsoneisthatLabeledtrainingsetisthebaseofmostmethodesofNLPhowevertheworkoflabelingthetrainingsetbyhandiscostlyitneedsmanypeoplewithstrongexpertknowledgeworkinghardly.Theotheriswiththeadventoftheinformatio

5、neraandthedevelopmentoftheInternetthecontentofInternetincreaseswithexponentialspeedwecangettheserawdatafreelyandusethemintheNLPresearch.InthispaperwebuildaresearchworkontherecognitionofChinesechunkwiththeCotrainingmethod

6、.WegivethedefinitionofChineseChunkthendiscussformalizeddefinitionofCotrainingalgorithmunderthePACframework.Firstlywedefinethetwo“views“ofexamplesbychoosingtwoClassifiersbasedondifferentalgorithmtheory.Andthenweproposedae

7、xampleselectionmethodbasedontheconsistenceusingtwoclassifiers:TransductiveHMMandfhTBLtocombineaclassificationsystemtoperformtheChineseTextchunkingtaskwiththesmallscalelabledChinesetreebankandlargescaleunlabledChinesecorp

8、us.TheresultwerecomparedwiththeselftrainingresulttheresultofthenonCotrainingexperimentinwhichweonlyusedthesmallscaleChinesetreebankastrainingdataanduseoneclassifier(TransductiveHMMorfhTBL)torecognizetheChinesechunk.Theim