基于RBFN——项目聚类的协同过滤推荐算法研究

Research on Collaborative Fltering Recommendation Algorithm Based on Item Clustering and RBFN

高维数据的稀疏性问题是降低协同过滤技术推荐质量的主要原因之一。提出了基于径向基函数网络(RBFN)——项目聚类的算法来降低数据的稀疏性,应用径向基函数网络(RBFN)处理高维稀疏数据得到一个完整的矩阵,应用基于项目聚类的协同过滤推荐算法产生推荐。实验结果表明,本算法比其他算法能更好处理协同过滤中的稀疏性问题。

The sparsity of high dimensional data is a major reason to reduce the recommend quality of collaborative filtering. This paper comes up with the RBFN - based and item - based clustering hybrid algorithm to reduce the data sparsity. we apply RBFN to get a complete matrix and employ a collaborative filtering recommendation algorithm based on item clustering to obtain recommendations. The experiment results suggest that the new algorithm can better handle the matrix sparsity problem.