問答社區(qū)中的問題與答案推薦機制研究與實現(xiàn).pdf

1、浙江大學計算機科學與技術(shù)學院碩士學位論文問答社區(qū)中的問題與答案推薦機制研究與實現(xiàn)姓名：曲明成申請學位級別：碩士專業(yè)：計算機應用技術(shù)指導教師：卜佳俊王燦20100301AbstractUserInteractiveQuestionAnswering(QA)communitiessuchasYahoo!AnswersandBaiduZhidaoaregrowinginpopularityHoweverastheseQAsitesalways

2、havethousandsofnewquestionsposteddailyitisdifficultforuserstofindthequestionsthatinterestthemConsequentlythismaydelaytheansweringofnewquestionsMeanwhile，asthenumberofcandidateanswersincreasing，theaskermayhavedifficultyin

3、choosingthebestonefromthosewithunevenqualityInthispaperwestudythequestionandanswerrecommendationmechanismtohelptheaskerandanswererseekinformationandenhancetheknowledgesharingactivitieswithinquestionansweringcommunitiesQu

4、estionrecommendationtechniqueshelpuserslocateinterestingquestionsandexpeditetheansweringofnewquestionsWebelieveusersmayselectquestionsaccordingtOtheirownpreferencesThusinthispaperweadopttheProbabilisticLatentSemanticAnal

5、ysis(PLSA)modeltopresentusers，preferencedistributionaccordingtousers’previousansweringhistorythusgeneratingrecommendingquestionlistsTohelpaskersfindthebestanswers，answerrecommendationtechniquesrankcandidateanswersautomat

6、icallyTherecommendationisconductedbasedOi“1thecontentsimilarityanduserauthorityWesummarizetherelationshipbetweenusersaslinkstructure，andadoptthePageRankalgorithmforestimatinguserauthorityThesimilaritycalculationconsiders

7、boththesimilaritybetweenquestionsandanswers，andthatbetweenaskersandanswersTheexperimentalresultsshowourtopicmodelingbasedquestionrecommendationapproachcancaptureusers’preferenceandrecommendquestionseffectivelyExperimenta

8、lresultsofanswerrecommendationconcludetheeffectivenessofconsideringboththesireilaritybetweenquestionsandanswers，andthatbetweenaskersandanswersKeywords：QuestionAnsweringCommunities，QuestionRecommendation，AnswerRecommendat