CAJ | 학술논문

结合社会网络分析的推荐方法研究已成为热点。电子商务中用户的动态行为异常丰富,隐含了用户的关联关系,利用这些信息进行商品推荐是个新研究思路。分析电子商务系统中用户动态行为关联关系及用户间明确好友关系形成复杂隐性社会网络,将社团划分算法应用到该网络中,则社团内部用户联系紧密且具有更相似的消费偏好,据此设计了电子商务中社团内部的推荐方法,应用R语言进行了算法的验证并与传统的协同过滤算法进行比较。实验表明,该推荐算法提高了推荐的质量,缓解了传统推荐算法中数据稀疏性及冷启动问题等。
결합사회망락분석적추천방법연구이성위열점。전자상무중용호적동태행위이상봉부,은함료용호적관련관계,이용저사신식진행상품추천시개신연구사로。분석전자상무계통중용호동태행위관련관계급용호간명학호우관계형성복잡은성사회망락,장사단화분산법응용도해망락중,칙사단내부용호련계긴밀차구유경상사적소비편호,거차설계료전자상무중사단내부적추천방법,응용R어언진행료산법적험증병여전통적협동과려산법진행비교。실험표명,해추천산법제고료추천적질량,완해료전통추천산법중수거희소성급랭계동문제등。
Recommended method combined with social network analysis has become a hot spot. The dynamic behavior of nsers is unusually rich in e - commerce implied the user＇s relationship and with the use of the information for recommendation is a new research idea. According to this can construct a complex recessive social network by the user dynamic behavior relationship and clear relationship between users of the e - commerce and using conmaunity partition algorithm on it, the internal users are linked closely and have more similar consumption preference, and design a recommended method based on community partition. Using R language for the validation of the proposed algorithm and comparison with the traditional collaborative filtering algorithm. Experiments show that the recommendation algorithm improves the quality of the recommendation and alleviates the data sparseness and cold start problem in traditional recommendation algorithm.