摘要
综合用户背景与资源特点,基于用户的协同过滤更适合高校图书馆在信息推荐中的应用。对其由于馆藏数字资源空间增大而导致推荐系统性能下降以及数据稀疏性问题,提出一种用户意图聚类的方法。通过运用K-m eans算法,对资源类别的意图特征值相似用户进行聚类,来提高推荐的实时性,降低数据稀疏性对信息推荐造成的影响。实验结果表明,基于用户意图聚类的协同过滤算法能有效提高推荐质量。
Integrating users＇ background with resources＇ characteristics, the user-based collaborative filtering fits university libraries in information recommendation better. As for the lower performance of the recommendation system and the sparsity of the data caused by the increased space of library digital resources, this paper proposes a user intention-clustering method. By the use of the K-means algorithm, the users having similar intent eigenvalue of resources＇ types are clustered to increase the real time of recommendation and decrease the influence of data sparsity on information recommendation. The experimental results indicate that the collaborative filtering algorithm based on user intention clustering can improve the quality of recommendation effectively.
出处
《情报理论与实践》
CSSCI
北大核心
2011年第6期116-119,共4页
Information Studies:Theory & Application
基金
上海应用技术学院社会科学基金项目"数字资源检索中的LibSuggest模式及其应用研究"的成果
项目编号:SJ2010-04
关键词
数字资源
聚类分析
用户
协同过滤
digital resources
clustering analysis
user
collaborative filtering
