A hybrid recommender system using multi layer perceptron neural network
Didar Divani Sanandaj, Sasan Hossein Alizadeh · 2018
Recommender systems try to personalized preferences, obviously this predicting based on previous users taste. Many RS suffer from Cold-start problem, it pertains to the issue which the system cannot invoke any reasoning for users whom they have not yet accumulated proper information. In this paper, we will propose a hybrid recommender system based on collaborative filtering (CF) techniques and content-based filtering (CBF) which combined by artificial neural network (ANN) to be an appropriate model for both cold-start and ordinary users. In our proposed method, Mutual Information techniques help us to select the best property to make our model. The comparison is performed on MovieLens and Netflix datasets. Our results show that our prediction results are more accurate than other methods with the same dataset.