Recommender engine.pdf

《Recommender engine.pdf》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

1、RWTHAachenUniversityUniversityofBonnFraunhoferFITE-CommerceSeminarWT08/09RecommenderEnginesSeminarPaperThomasHess(289222)February1,2009AbstractRecommenderenginesareusedbymoreandmoree-commercebusinessestohelpcon-sumersfindingproductstheyareinterestedin.Thepaperdescrib

2、eswhatrecommenderenginesareandwhatroletheyplayine-commerce.Recommenderenginesusevarioustechniquesthatusedif-ferentknowledgesourcestomakerecommendations.Thepaperexplainsthesetechniquesandtheirstrengthsandweaknesses.Someofthecommonissuesthatrecommendersystemsfacearedi

3、scussedandpossiblesolutionspresented.Concludingexamplesofrecommenderenginesine-commercearedescribed.Itisshownwhattechniquestheyuseandhowthee-businessesutilizerecommendationsontheirwebsites.Contents1Introduction52RecommenderTechniques62.1Non-PersonalizedRecommendatio

4、n..........................62.2DemographicRecommendation............................72.3Content-BasedRecommendation............................82.4CollaborativeFiltering.................................92.4.1User-BasedApproach.............................102.4.2Item

5、-BasedApproach.............................102.4.3Model-BasedApproach............................122.5HybridApproaches...................................133IssuesAndSolutions143.1DataCollection.....................................143.2ColdStart......................

6、..................143.3Stabilityvs.Plasticity..................................153.4Sparsity.........................................153.5Performance&Scalability...............................163.6UserInputConsistency.................................173.7Privacy....

7、.....................................174RecommenderEngineExamples194.1ChoiceStream......................................204.2Amazon.com......................................224.3Digg...........................................295Conclusion363ListofFigures2.1Knowledge

8、SourcesofRecommenderEngines.....................62.2Non-PersonalizedRecommendation..........................72.3DemographicRecommendation.