基于案例推理的过滤算法及智能信息推荐系统

智能信息推荐系统能够通过用户偏好,利用信息过滤算法主动剔除无关信息。该文提出了一个基于案例推理的职能信息推荐系统的架构,主要包含数据层、过滤层和结果展示层3个功能层次。设计了一个基于案例推理的过滤算法,将用户对文本的评价定义为案例;采用归一化的Euclidean距离,计算用户之间的相似度。在一个公共数据集上进行了不同过滤算法的reca ll值对比实验。结果表明,采用案例推理技术,对协同过滤的准确度有一定程度的改善。

CBR filtering algorithm and intelligent information recommending system

Intelligent information recommending systems can kick out user-useless information using filtering algorithms and user's profile.This paper describes the architecture of an intelligent information recommending system which contains three functional levels,data level,filtering level,and result expressing level.A CBR(case-based reasoning) filtering algorithm was designed.The use's evaluations to the documents were defined as cases.Euclidean distance was used to compute the users' similarity.Finally experiments were made on a common data set using different filtering algorithms.The recall values were compared and analyzed.The results show that the CBR can improve the excitability of the collaborative filtering system.