基于产品分类的协同过滤算法应用研究

摘要

In order to remedy the sparsity and scalability weaknesses and improve the recommendation quality, based on conventional collaborative filtering algorithms, product taxonomy and web usage mining technologies are introduced. After elaborating the algorithm in detail, with experimental data the algorithm' s recommendation performance is validated. Experiment results show that the performance of collaborative filtering algorithm is significantly improved and meanwhile the sparsity and scalability weaknesses are greatly remedied by the utilization of product taxonomy and web usage mining technologies.%为了解决稀疏性问题和可扩展性问题,提高推荐的质量,在传统协同过滤算法的基础上,引入产品分类技术与Web使用挖掘技术.在详细阐述算法的基础上,通过实验数据验证该算法的推荐性能.实验结果表明,引入产品分类和Web使用挖掘技术后,协同过滤算法的性能有了显著的提高,很好地改善了其稀疏性问题和可扩展性问题.