基于语义相似性的资源协同过滤技术研究  

A Study on Item-Based Collaborative Filtering Algorithm Using Semantic Similarity

作　　者：崔林[1] 宋瀚涛[1] 陆玉昌[2] 

机构地区：[1]北京理工大学信息科学技术学院计算机科学工程系 [2]清华大学计算机科学与技术系,北京100084

出　　处：《北京理工大学学报》2005年第5期402-405,共4页Transactions of Beijing Institute of Technology

基　　金：国家"九七三"计划项目(G1998030414)

摘　　要：为解决协同过滤推荐系统中所存在的可扩展性、稀疏性等问题带来的推荐性能下降,提出新的基于资源语义知识协同过滤算法,算法综合考虑了资源语义和用户评价的影响,改善基于资源协同过滤算法性能.实验表明,基于资源语义的协同过滤算法相对于传统协同过滤算法可提高推荐性能.In a recommendation system based on collaborative filtering(CF), in order to resolve efficiently some problems such as the scalability and sparsity problems the quality of recommendation system will tend to be decreased dramatically. A new CF algorithm based on semantic knowledge of items is presented. The algorithm takes synthetically into account the influence of item semantic and user rating to enhance the item-based CF. Experimental results indicate that the algorithm can achieve better prediction accuracy and provide better recommendation results than with the traditional CF algorithms.

关 键 词：个性化 推荐系统 协同过滤 基于资源CF 语义相似性 

分 类 号：TP182[自动化与计算机技术—控制理论与控制工程]