基于邻居决策的协同过滤推荐算法

协同过滤技术应用于个性化推荐系统中，稀疏性问题和可扩展性问题成为亟需解决的问题。针对传统方法的不足，提出一种凭借邻居数做决策的方法，比较各个待测位置的用户邻居数和项目邻居数，由数量多的一方作预测，同时对预测值判定给出一种合理而有效的度量方法。实验结果表明，该方法能够提高推荐质量。

Collaborative Filtering Recommendation Algorithm Based on Neighbor Decision-making

Collaborative filtering has been applied in personalized recommendation system successfully,sparsity problem and scalability problem become two big problems which remain unresolved.To slove the problem of traditional method,this paper propose a decision-making method relying on the number of neighbors.The method compares the number of user’s neighbors and item’s neighbors in every unpredicted position,and chooses the bigger one to make predicting.In addition,a reasonable and effective measurement is put forward to judge predicting.Experimental result shows that the quality of recommendation is largely improved.