基于影响集的协作过滤推荐算法.pdf

ISSN RUXUEW 1000、9825，CODEN Journal ofSoftware，V01．18，No．7，July2007，PP．1685—1694 DOI：10．1360，josl81685 @2007Journal reserved． by ofSoftware．Allrights 基于影响集的协作过滤推荐算法宰 陈健1+，印鉴2 1(华南理工大学计算机科学与工程学院，广东广卅I510006) 2(中山大学计算科学系，广东广州510275) A on CollaborativeRecommendationBasedInfluenceSets Filtering Algorithm CHEN Jianl+，YINJian2 of and Science China of 5 1(SchoolComputer Engineering，SouthUniversityTechnology，Guangzhou10006，China) of Yat—Set 5 Science，Sun 10275，China) 2(DepartmentComputer University，Guangzhou +Corresponding Chen collaborativerecommendationbasedinfluence J，YinJ．A on sets．Journal filtering algorithm of Abstract：Thetraditionaluser-basedcollaborative oftensufferfromtwo filtering(CFlalgorithms important and becauseofits knearest CF memory—based problems：Scalabilitysparsity neighborquery todealwith havebeen the associatedwithuser-basedCF algorithms designed scalabilityproblems without recommendationor CF stillsufferfromthe sacrificing predictionaccuracy．However,item—basedalgorithms data aCFrecommendation sparsityproblems．Thispaperpresents algorithm，named filtering basedoninfluence isbasedohthe ofinfluencesetandisahot ininformationretrieval sets)，which concept topic