基于分页缓存模型的用户兴趣跟踪方法

J4 ›› 2012, Vol. 34 ›› Issue (10): 32-37.

基于分页缓存模型的用户兴趣跟踪方法

李志浩1，聂文汇1，成鹏2，张宇博2，阳智敏2

(1.武汉大学计算机学院，湖北武汉 430072;2.武汉大学软件工程国家重点实验室，湖北武汉 430072)

收稿日期:2012-04-25 修回日期:2012-07-10 出版日期:2012-10-25 发布日期:2012-10-22
基金资助:
国家973计划资助项目（2007CB310806）；NOKIA校企合作项目

User Interest Tracking Method Based on Paging Cache Model

LI Zhihao 1,NIE Wenhui 1,CHENG Peng 2 ,ZHANG Yubo2,YANG Zhimin2

(1.School of Computer,Wuhan University,Wuhan,430072;2.State Key Laboratory of Software Engineering,Wuhan University,Wuhan 430072,China)

Received:2012-04-25 Revised:2012-07-10 Online:2012-10-25 Published:2012-10-22

摘要/Abstract

摘要：

对智能推荐系统中用户兴趣跟踪问题的研究，传统方法如时间窗口、遗忘函数等在表征用户兴趣模型时均未考虑兴趣主题概念相关性，无法充分利用用户历史数据，导致兴趣跟踪不准确。因此，本文提出了基于分页缓存的用户兴趣表征模型，形成基于主题的用户多兴趣域结构，并提出了相应的兴趣迁移检测SIM算法，该算法引入序列熵差，表征兴趣迁移的整体特性。实验表明，与传统方法相比，本文提出的方法具有更低的兴趣平均绝对偏差，能够更准确地表征用户兴趣迁移，从而获得更好的推荐质量和效率。

关键词: 分页缓存, 兴趣迁移, 序列熵差, 兴趣更新

Abstract:

Aiming at tracking user’s interest in intelligent recommender systems, traditional methods such as Time Window, Forgotten Function, consider little in topic relevance when characterizing user’s interest model and could hardly make full use of historic data, leading to inaccurate in tracking user’s interest migration. Therefore, the paper proposes a new characteristic model based on paging cache, which forms user’s multiple interest domains classified by topics. This paper proposes the corresponding SIM algorithm which introduces sequence entropy difference to characterize the integral features of interest migration. Experimental results show that the proposed method provides lower mean absolute error of interest and precisely characterize user’s interest migration, so as to enhance the quality and efficiency in services.

Key words: paging cache;interest migration;sequence entropy difference;interest update

李志浩1，聂文汇1，成鹏2，张宇博2，阳智敏2. 基于分页缓存模型的用户兴趣跟踪方法[J]. J4, 2012, 34(10): 32-37.

LI Zhihao 1,NIE Wenhui 1,CHENG Peng 2 ,ZHANG Yubo2,YANG Zhimin2. User Interest Tracking Method Based on Paging Cache Model[J]. J4, 2012, 34(10): 32-37.