Recommendation based on rational inferences in collaborative filtering

In collaborative filtering, the existing memory-based methods make recommendations based on the overall consistency between two users or two items. The major concerns with these methods are: (1) they are sometimes being overly confident; (2) they are prone to disregard some useful information in the user profiles; (3) they often imply some untrustworthy inferences in making a prediction. This work investigates the drawbacks of these methods, and then proposes a collaborative filtering approach based on heuristic formulated inferences. The proposed approach is based on the fact that any two users may have some common interest genres as well as different ones. Different from most existing methods, this approach introduces a more reasonable similarity measure metric, considers users’ preferences and rating patterns, and promotes rational individual prediction, thus more comprehensively measures the relevance between user and item. Experimental results from two popular public datasets show that the proposed approach improves the prediction quality significantly over several other popular methods.