【摘要】 针对目前推荐算法面临的响应时间长、推荐不准确和冷启动的问题,本文以电影推荐系统为例提出了一种基于用户可信度聚类的协同过滤推荐模型,该模型将推荐过程分为离线和在线两部分。离线时,计算用户可信度,以此聚类,并将聚类信息储存在数据库中的聚类表里;在线时,寻找目标用户所属聚类,产生推荐。模型整体上减少了响应时间,提高了推荐准确率,同时解决了新用户冷启动问题。更多还原

【Abstract】 According to long response time,inaccurate recommendation and cold-start problem that present recommending algorithm faced,taking movie recommending system for example,this paper proposes a collaborative filtering recommending model based on user’s credibility clustering.The model splits computations into offline and online phases.Offline,it uses the result of user ’s credibility for clustering and then records the clustered information into a table which stored in database.The online phase projects target user into the cluster which it belongs to.At the end,the recommendation comes out.As a whole,the model reduces the response time and improves the accuracy of the recommended rate,while solves the new user’s cold-start problem.更多还原