Collaborative Filtering Recommendation Algorithm Based on Item Category and Time Factor

Abstract

The traditional collaborative filtering algorithm has problems, such as data sparsity, too much calculation and ignoring time factor. For these problems, this paper advances one new algorithm. And this algorithm based on item category and time factor. Firstly, we use time factor to weight the user's ratings, and fill user-item rating matrix. Then, we set up a user-category rating matrix, and use improved similarity calculation method to obtain the nearest neighbors and recommend. Finally, based on the MovieLens dataset experiment, this algorithm relieves the disadvantage of sparsity effectively, improves recommendation accuracy.