摘要
分析了现有文章推荐系统中基于关键词向量的用户模型表示方法存在的不足,提出了基于聚类兴趣点的用户模型表示方法。该方法可通过文章聚类形成兴趣点。由于传统的基于划分的聚类算法存在的不足,提出了基于复杂网络特征的文章聚类算法。实验结果表明该用户模型的表示方法较好地反映了用户多方面的兴趣,提高了文章推荐系统的性能。
After analyzing the disadvantages of the user profile based on-keywords vector in the existing document recommendation system, a novel representation of user profile based on clustering was proposed. The representation firstly clustered the documents into clusters. Because of the disadvantage of the traditional partitioned clustering algorithm, a novel document clustering algorithm based on complex networks featttre was presented. Experimental results show the representation of user profile proposed can represent user multi - interests better and improves the perforrnance of document recommendation system greatly.
出处
《计算机技术与发展》
2007年第1期4-5,48,共3页
Computer Technology and Development
基金
安徽省自然科学基金项目(2004kj011)
安徽省高校青年教师基金项目(2006jq1040)
关键词
聚类
复杂网络
推荐系统
用户模型
clustering
complex networks
recommendation system
user profile
