摘要
传统的协作筛方法存在的主要问题是需要人为地提供评价,该文对其进行了改进,根据用户事务模式自动获取用户评价,构建评价矩阵。针对现有个性化信息推荐系统存在的问题,将Agent应用到电子商务个性化信息推荐,引入设计的协作筛方法,设计了基于Agent的电子商务协作推荐系统(ECCRS)。ECCRS是基于服务器端的电子商务协作推荐系统,它考虑了用户访问页面的时间特性,不需要用户注册,在推荐时考虑了页面的新颖性,同时实现了离线处理、增量更新与在线推荐的结合。实验结果表明,ECCRS采用的推荐方法是有效的。
The paper improves the traditional collaborative filtering method which needs artificial evaluation. The advanced method can gain evaluation value automatically through user transaction pattern and establish evaluation matrix. Aiming at problems of current personalized information recommendation systems, this paper applies Agent and designs collaborative filtering method to E-commerce personalized information recommendation, and designed E-commerce collaborative recommendation system based on Agent, namely, ECCRS, which is based on server. ECCRS considers time of page calling and novelty of page. It can supply recommendation service for unregister users and combined off-line processing, incremental updating and on-line recommendation. The experimental results indicate that the recommendation method of ECCRS is feasible.
出处
《计算机工程》
CAS
CSCD
北大核心
2007年第9期216-218,共3页
Computer Engineering
基金
重庆市科委科技攻关基金资助项目(CSTC.2004AA2001-8277-9)
重庆工学院青年基金资助项目
关键词
电子商务
智能体
协作推荐
个性化主动信息服务
E-commerce
Agent
Collaborative recommendation
Personalized active information service
