摘要
为了解决WWW上的“信息过载”和“资源迷向”问题,该文提出了基于关联规则挖掘的个性化智能推荐服务。个性化智能推荐服务系统包括两个主要部分:离线部分和在线部分,在离线方式下,执行对WEB服务器的访问log文件的分析挖掘,获取用户事务模式,再采用支持度过滤方法获取频繁的用户事务模式,然后,生成聚集树。在在线方式下,针对当前滑窗的用户访问操作路径,采用基于聚集树的关联规则挖掘,获取匹配当前滑窗的用户访问操作路径的关联规则集,生成推荐的候选集。实现在线个性化智能推荐服务。试验结果显示,该文提出的方法是有效的和可行的。
This paper proposes an intelligent service method on personalized recomme ndation,which is based on associ-ation rules mining for Internet.To alleviate the phenomena of'information overload'and'information bewilderment 'in Inte rnet environment ,the overall process can be divided into two components:offl ine part and online part.In offline,WEB mining tasks can be executed in the lo gs of WEB service resulting in a user transaction file,and the frequent user t ransaction patterns are extracted by filtering with thresholds of support again ,afterwards,constructing aggregating tree of user sessions.In online,the can didate URLs for recommendation can be determined by matching association rules in the aggregating tree with the current active session for the intelligent se rvices of personalization recommendation.The experiments demonstrate that the a pproach is applicable and effective.
出处
《计算机工程与应用》
CSCD
北大核心
2002年第11期200-204,229,共6页
Computer Engineering and Applications
基金
教育部博士生专项基金资助(编号:1999035808)
