摘要
群体推荐是个性化推荐领域一个新的研究热点,其与传统个体推荐的重要区别之一,是需要群体成员进行协商或谈判,以提高群体满意度。针对该问题,提出了基于案例推理和协商的群体推荐算法,根据群体成员对项目的历史评价作为知识库,从群体用户角度出发进行协商或谈判,使用多Agent系统模拟群体用户在选择推荐项目问题上的协商过程,最终达成一致并进行推荐,推荐完成后通过用户反馈对群体成员的知识库进行及时更新。选取MovieLens数据库进行试验评价,结果表明,文中算法的推荐质量明显优于对比算法。
Group recommendation system （GRS） has become the focus in the research of personalized recommendation system, one of the main difference between it and the traditional individual recommendation system is that GRS requires consultation or negotiation among the members of the group, which aims to increase the group＇s whole satisfaction. To solve the problem, this paper proposes the group recommendation algorithm based on case-based reasoning and negotiation, which mainly builds a repository according to the historical ratings to the recommended items, uses multi-Agents system to simulate the negotiation process from user＇s perspective, and finally reaches an agreement with the result, meanwhile updates the repository with user＇s feedback. We find the group recommendation algorithm based on case-based reasoning and negotiation is more effective in the test.
出处
《系统工程》
CSSCI
CSCD
北大核心
2013年第11期93-98,共6页
Systems Engineering
基金
国家自然科学基金资助项目(71271147)
