Abstract: This paper proposes a new approach for constructing user profiles based on user feedback. In contrast to earlier approaches, which typically represents profiles as a single weighted interest vector, it represents user profiles as a single integrated vector and a multiple interest vectors, and takes feature selection into the system to reach higher performance. This flexible approach allows the profile to more accurately represent complex user interests and faster to locate for information of interest.

Key words: Information filtering; Adaptive; User profile; Feature selection

摘要： 提出了一种新的、基于用户反馈的、构建用户模板的方法，它不同于传统的方法将用户兴趣当作单模板来处理。引入了特征选择技术，能够对用户模板进行动态调整，并构建了自适应的单模板和多模板系统，能够更准确地表示用户兴趣和定位相关信息。

关键词: 信息过滤；自适应；用户模板；特征选择