A collaborative filtering recommendation algorithm based on multi-dimensional data filling
Meiqing Song · 2016
This paper proposes a collaborative filtering recommendation algorithm based on multi-dimensional data filling aiming at data sparsity of collaborative filtering. In order to find the dimensions which not only make data collecting more convenient, but also distinguish the user's interests prominently, this algorithm evaluates the user's dimensions. Then user-item rating matrix is filled with multi-dimensional data in the basis of the chosen dimensions. In this way, the degree of data sparsity of rating data is decreased and the similarity judgment is provided with more data. The results obtained by contrast experiment present that the algorithm proposed in this paper has higher efficiency of recommendation than traditional algorithm.