计算机科学与探索 ›› 2012, Vol. 6 ›› Issue (4): 333-342.DOI: 10.3778/j.issn.1673-9418.2012.04.006

• 学术研究 • 上一篇    下一篇

新的基于日志分析的用户个性化模型

杨 艳,邱艳丽   

  1. 1. 黑龙江大学 计算机科学技术学院,哈尔滨 150080
    2. 黑龙江大学 数据库与并行计算重点实验室,哈尔滨 150080
  • 出版日期:2012-04-01

New Log-Based Personalized User Model

YANG Yan, QIU Yanli   

  1. 1. School of Computer Science and Technology, Heilongjiang University, Harbin 150080, China
    2. Key Laboratory of Database and Parallel Computing, Heilongjiang University, Harbin 150080, China
  • Online:2012-04-01

摘要: 作为个性化服务技术的核心,用户模型的质量关系到个性化服务的质量。目前的用户模型大多只考虑用户的显式信息或隐式信息,很少同时考虑两者,使得检索质量不如人意。提出了一种新的基于日志分析的用户个性化模型,结合了传统的显式建模和隐式建模的优点,把显式个性化信息和隐式个性化信息通过两层树状结构结合起来。模型同时考虑了用户历史信息的长短期划分,以及检索系统返回结果的顺序和用户对结果页面的点击顺序。实验结果表明,基于该用户模型的个性化检索效果与原有检索系统的检索效果相比有显著提高。

null

关键词: 计算机软件, 用户模型, 显式反馈, 隐式反馈, 个性化, 日志

Abstract: As the core of personalization, the quality of personalized service is tightly related with personalized user model. Current user models take into account only explicit information or implicit information of users, rarely combine both. As a result, the result of the personalized retrieval does not satisfy the users. This paper proposes a new log-based personalized user model, which combines explicit feedback and implicit feedback through a two-layer tree. The model also divides user information into long-term information and short-term information, and considers the order of the results returned by information retrieval system and the order of user clicks on the results. Experimental results show that the efficiency of the personalized retrieval based on the proposed log-based model is better than general retrieval system.

Key words: computer software, user model, explicit feedback, implicit feedback, personalization, log