在协同过滤中结合奇异值分解与最近邻方法

协同过滤是一种减小信息过载的常用方法，但是它有三方面的限制，即准确性、数据稀疏性和可扩展性。提出一种新的协同过滤算法来解决数据稀疏性的问题，利用奇异值分解法的结果来进行邻居选择，然后采用最近邻方法来得到未打分项目的预测值。在EachMovie 数据库集上的试验结果表明该算法在数据稀疏时算法的准确性超过普通的Pearson算法和奇异值分解算法。

Combining Singular Value Decomposition and Neighbor based Method in Collaborative Filtering

Collaborative filtering is becoming a popular technique for reducing information overload.However,it has three major limitations,accuracy,data sparsity and scalability.We propose a new collaborative filtering algorithm to solve the problem of data sparsity.We utilize the results of singular value decomposition for neighbors selecting,then use the neighborhood-based method to produce the prediction of unrated items.Our experimental results on EachMovie dataset show that the algorithm outperforms the conventional neighborhood-based method and SVD method when the available ratings are sparse.