基于部分计值的信息个人化智能Agent系统

Intelligent Information Personalization Agent System Based on Partial Evaluation

提出了一个基于部分计值技术的信息个人化系统PIPEAgent。部分计值理论源于程序转换中如何在给定部分输入的情况下，对程序完成尽可能多的运算从而对程序进行优化这一问题的探讨，PIPEAgent将部分计值理论与信息个人化相结合，以PIPE(Personalization is Partial Evaluation)理论为基础，结合机器学习技术并采用智能人机交互的Agent构架。该系统已应用于Web信息检索、电子商务应用。

This paper describes the architecture of PIPEAgent which is an intelligent information personalization system based on partial evaluation. The theory of partial evaluation originated from the discussion about how to finish the program computation under some known inputs as much as possible. Since the redundant part of the program is removed during the optimization, the efficiency of it will be improved. Based on partial evaluation technology, PIPEAgent optimizes the Web organization and makes its information personalized to various users.