An ensemble method for enhancing collaborative filtering recommender systems by MLP neural network based on trust propagation

Recommender systems have been shown to be an effective way to assist users to find what they need in unlimited amount of data in the internet. In this way, by personalizing information and acting like a filtering system, they are a subclass of decision support systems. One of the best techniques for improving quality of recommendations is using trust relation between users, but sparsity is a critical problem of trust aware recommender systems; in this paper we use ensemble method that combines trust propagation results as a solution for solving the sparsity problem and Multi Layer Perceptron neural network to provide more accurate recommendations. In order to the test the proposed method we used Extended Epinions datasets, and the results confirm improvement on sparsity and on accuracy.

کلیدواژه ها:

Recommender Systems ، Neural Network ، Trust Propagation