A pattern mining approach to enhance the accuracy of collaborative filtering in sparse data domains

Recommender systems seek to find the interesting items by filtering out the worthless items. Collaborative filtering is one of the most successful recommendation approaches. It typically associates a user with a group of like-minded users based on their preferences over all the items and recommends the items which are welcomed by others in the group to the user. But, many challenges like sparsity and computational issues still arise. In this paper, to overcome these challenges, we propose a novel method to find the neighbor users based on the users’ interest patterns. The main idea is that users who are interested in the same set of items share similar interest patterns. Therefore, the non-redundant item subspaces are extracted to indicate the different patterns of interest. Then, a user’s tree structure is created based on the patterns he has in common with the active user. Moreover, a novel recommendation method is presented to predict a new rating value for unseen items. Experimental results on the Movielens and the Jester datasets show that in most cases, the proposed method gains better results than already widely used methods.