基于信任的电子商务个性化推荐关键问题研究

《基于信任的电子商务个性化推荐关键问题研究》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、江西财经大学博士学位论文基于信任的电子商务个性化推荐关键问题研究姓名：张富国申请学位级别：博士专业：管理科学与工程指导教师：徐升华20090501(3)提出了基于信任的推荐多样性算法，该算法通过选择主题多样的信任邻居来平衡推荐结果的准确性和多样性。并通过一系列的实验结果表明，该算法能有效地提高推荐的多样性。(4)在对信任传递可能引起的恶意攻击形式进行分析的基础上，提出了用数-据起源法来追踪恶意用户，进而把恶意用户限制为不信任用户的防御方法。关键词：信任；个性化推荐；协同过滤：信任传递；电子商务AbstractWiththerap

2、id．popularizationofIntemet，virtualshoppingenvironmentofe-commercenotonlyprovidenewdevelopmentchancefortheenterprises，butalsoputforwardhowtohandlethechallengeofinformationoverloadofwebproducts．Asameansofpersonalizedservice，recommendersystemsCanrecommendinterestingitem

3、sforusers，assistuserstomakedecision，andbecomeallgoodassistantforusers’shoppingonline．Sincetheemergenceofthefirstrecommendersysteminthe1990s，remarkableprogressinthisfieldhasbeenachieved，buttherestillexistssomeissueshardtoovercomeintraditionalcollaborativefilteringtech

4、nology．Inthesituationoftherapiddevelopmentofwebsocialnetworks,becauseofincorporatingtrustmechanismintotraditionalcollaborativefilteringrecommendersystems，trust-basedrecommendersystemsCanimproveorovercometheselimitationseffectively,andbecomeoneofthemostimportantresear

5、chprojects．Thisthesisexploredandresearchedsomekeyissuesoftrust—basedpersonalizedrecommendersystemsine-Commerce．Firstly,itgaveanoverviewofthetheoreticalresearchanddevelopmentonpersonalizedrecommendationtodate．Theadvantagesof‘collaborativefilteringtechnologyandresearch

6、challengeswereanalyzed．Secondly,fromthedefinitionandpropertiesofnetworktrust，thecharacteristicofthetrustevaluationmodelswereanalyzedandcompared．Thirdly,ageneralmodeloftrust-basedpersonalizedrecommendersystemandsystemframeworkwaspresented．Thentheauthorgavetheresearchw

7、orkontrust·basedpersonalizedrecommendationalgorithmforuser’smultipleinterests，diversityofrecommendationlistitemsandrecommendationattackoftrust-basedrecommendersystem．。Theinnovationalworksofthisthesisareasfollowing：(1)Basedonthebasicmodaloftraditionalcollaborativefilt

8、eringsystemandthecharacteristicofwebsocialnetwork，theauthorpresentedageneralmodaloftrust—basede-commercepersonalizedrecommendersyst