面向個性化web新聞服務的層次型用戶興趣建模研究

1、合肥工業(yè)大學碩士學位論文面向個性化Web新聞服務的層次型用戶興趣建模研究姓名：李世瀛申請學位級別：碩士專業(yè)：計算機應用技術指導教師：胡學鋼；吳共慶201204HierarchicalUserInterestModelingforPersonalizedWebNewsServiceABSTRACTWiththerapiddeVelopmentofInternettechnology，webdata1s1ncreaslngexplosive

2、lyInthevastamountsofdata，howtoaccesstheinterestlngIntormatlonfastandefficientlygraduallybecomesthefbcusofusers’attentionWebDersonalizedservicesolvesthecontradictionbetweentherapidgrowthofwebinfj≯nTlationandtherelativelys

3、implemeanstoaccessln士onnatlontoacertalnextentWebinfomationserviceprovidersattractusersandimproVeusersatisfactionbycontinuouslyimproVingthequalityofpersonalizedserViceUserinterestmodelingisthecoreofpersonalizedserVicesIti

4、sappliedinthefj：eldsofinformationretrieval，datamining，ecommerceandpersonanzedrecommendationtoimprovethequalityofinfomationservicesMostoftraditionaluserinterestmodelsarebuiltonVSM(VectorSpaceModel)usingkeywordsastheuserin

5、terestHowever，thesemodelsnotonlyignorethehierarchicalgranularityrelationsbetweenkeywords，butalsoignoretheuseofdomainknowledgetoidentifythespecificconceptsortopicsofuserinterestsThus，itisdifficulttoexpresstheuserinteresta

6、ccuratelyandreasonablyintheuserinterestmodelingMotivatedbytheaboveproblem，usingpersonalizedwebnewsserVlceasresearchbackground，westudyhierarchicaluserinterestmodelingWeproposeaGraphPartitionbasedChinesePhrasesHierarchical

7、C1usteringalgorithmcalledGCPHCItorganizestheuserinterestinahierarchicaltree，designsaHowNetbasedMaximumMatchingMappingmethodcalledHNM3tomaptheuserinteresttothetopicsof0DP，andbuildsahierarchicaluserinterestmodel1abeledwith

8、thetopicforeachclusternodeThemainworkofthisdissertationisasfollows：f11InthecontextthatwebpersonalizedserViceincreasinglyneedstobemoreintelligent，thisdissertationusesadivisivehierarchicalclusteringalgorithmtobuildahierarc

9、hicaluserinterestmodelforusersinterestedwebcontent，whichishelpfulforwebpersonalizedserVice；(2)Fortheproblemofidenti母ingthespecificuserinteresttoplcs1nuserinterestmodel，thisdissertationusesHowNetandODPtomapeachhierarchica