A Computational Model for Trust-Based Collaborative Filtering - An Empirical Study on Hotel Recommendations

The inherent weakness of the data on user ratings collected from web, such as sparsity and cold-start, has limited the data analysis capability and prediction accuracy in recommender systems (RS). To alleviate this problem, trust has been incorporated in collaborative filtering (CF) approaches with encouraging experimental results. In this paper, we propose a computational model for trust-based CF combined with k-means clustering, k-nearest neighbor (kNN) and three different methods to infer trust, based on a detailed data analysis of hotel dataset collected from Booking.com. We apply this model on users’ ratings of hotels and show its feasibility by comparing the testing results with conventional CF algorithm using evaluation metrics Mean Absolute Error (MAE) and prediction coverage. Our experimental results indicate that the use of trust can improve prediction accuracy if the definition of trust is reasonable enough.