Personalized Context-Aware Rating Prediction Model and Recommendation Approach Based on Neural Network
Jiang Feng · 2010
Recent years, context has been identified as an important factor in recommender systems. Great contributions have been done for context-aware collaborative filtering recommendation approaches, but the contextual parameters in current approaches have same weights for all users. In the paper a recommendation approach based on BP neural network is proposed to learn a personal context-aware rating prediction model for each user. Each input unit in the model represents a contextual parameter and the weights of neural units are different for every user. Finally, we evaluate experimentally our approach and compare it to context-based collaborative filtering and Slope One. The experimental results show our algorithm out performs Slope One and traditional context-aware collaborative filtering.