基于用户-主题关联挖掘的虚拟社区推荐方法研究

Research on Virtual Community Recommendation Method Based on User-topic Association Mining

[目的/意义]针对现有的虚拟社区推荐方法缺乏兼顾推荐准确性和新颖性的问题,将数据挖掘技术与信息推荐方法相结合,提出了基于用户-主题关联挖掘的虚拟社区推荐方法。[方法/过程]该方法通过构建用户-用户相似度矩阵、社区-社区主题距离矩阵、基于矩阵分解的智能推荐等过程,使得推荐结果能在保证高准确性的前提下,兼顾推荐的新颖性。[结果/结论]实验结果表明,该方法取得了理想的预期结果,推荐效果既能保证准确性,又能体现新颖性。

[ Purpose/Significance I Aimed at solving the problem of lacking accuracy and novelty of the existing virtual community rec- ommendation method, this paper combined the data mining technology with information recommendation method and proposed the recom- mendation method of virtual community based on the user-topic association mining. [ Method/Process ] The method can not only make the recommendation result more reasonable, but also ensures its novelty by constructing usel-user similarity matrix, community-community topic distance matrix and intelligent recommendation based on matrix decomposition. [ Result/Conclusion] The experimental results show that the method has obtained the ideal expected recommendation results, not only guaranteeing the accuracy, but also reflecting the novelty.