基于半監(jiān)督學(xué)習(xí)的關(guān)鍵詞抽取.pdf

1、廈門大學(xué)碩士學(xué)位論文基于半監(jiān)督學(xué)習(xí)的關(guān)鍵詞抽取姓名：高燕申請學(xué)位級別：碩士專業(yè)：模式識別與智能系統(tǒng)指導(dǎo)教師：史曉東201207AbstractAbstractInrecentyears，BloghasbeendevelopingrapidlyasanewkindofinformationtransmissionandinteractivewayAsthenumberofusersincrease，theinformationisalso

2、increaserapidly。SouserssearchandpositioningtheinformationthattheyinterestedisalsodifficultTheKeywordscansupplyaconcisesummaryforanarticletoreadersItCanmakepositioninginformationmoresimplyintheInformationRetrievalKeywords

3、Extraction，alsoknownasAutomaticIndexingwhichusethecomputerstoextractthekeywords，andCancontainthethemeinformationofanarticleThispaperregardthetagsofblogasthekeywords，thendosomeresearchtokeywordsextractionforblog，according

4、tothestructureoftheblogandsomefeaturesoftags1WithrespecttoKeywordsExtraction，ithasbeenconsideredasabinaryclassificationproblem：an—keywordandkeywordOurworkishowtochoosethekeywordsfromthesetofcandidate；2Accordingtesearchin

5、gthefeaturesofthetagsofblog，wefindthatthetagsofbloghavesomeobvioustendencyoftopic，andsomehabitofauthorSoweproposedtwonewfeaturesofauthorandthetopicWeaddedthetwofeaturestothefeaturevectorandimprovethekeywordsextractioneff

6、ect；3Consideringthelackofthecorpusthathastagged，wegiveamethodwhichextractingthekeywordsbasedonsemi—supervisedlearningWehopethismethodCansolvetheproblemofkeywordsextractionusenotonlythesetofcorpusthathastagged，butalsothes

7、etofcorpusthathasnottagged；andwetrytotwokindofmethodbasedonsemisupervised1e鋤in爭一themethodofSVMbasedsemisupervisedlearningandlabelpropagatio仆一tosolvetheproblemofkeywordsextraction；4Wediscussedtheproblemofimbalancedatainth