In this paper we introduce the first application of the Belief Propagation(BP) algorithm in the design of recommender systems. We formulate therecommendation problem as an inference problem and aim to compute the marginalprobability distributions of the variables which represent the ratings to bepredicted. However, computing these marginal probability functions iscomputationally prohibitive for large-scale systems. Therefore, we utilize theBP algorithm to efficiently compute these functions. Recommendations for eachactive user are then iteratively computed by probabilistic message passing. Asopposed to the previous recommender algorithms, BPRS does not require solvingthe recommendation problem for all the users if it wishes to update therecommendations for only a single active. Further, BPRS computes therecommendations for each user with linear complexity and without requiring atraining period. Via computer simulations (using the 100K MovieLens dataset),we verify that BPRS iteratively reduces the error in the predicted ratings ofthe users until it converges. Finally, we confirm that BPRS is comparable tothe state of art methods such as Correlation-based neighborhood model (CorNgbr)and Singular Value Decomposition (SVD) in terms of rating and precisionaccuracy. Therefore, we believe that the BP-based recommendation algorithm is anew promising approach which offers a significant advantage on scalabilitywhile providing competitive accuracy for the recommender systems.
展开▼
