108 страниц. 2011 год. LAP Lambert Academic Publishing Recommender systems aim to suggest relevant items that are likely to be of interest to the users using a variety of information resources such as user profiles, trust information and users past predictions. However, typical recommender systems suffer from poor scalability, generating incomprehensible and not useful recommendations and data sparsity problem.In this work, we have proposed a probabilistic matrix factorization based local trust boosted recommendation system which handles data sparsity, scalability and understandability problems. The method utilizes the implicit trust in the review ratings of users. The experiments conducted on Epinions.com dataset showed that our method compares favorably with the methods in the literature.In the scope of this work, we have analyzed the effect of latent vector initialization in matrix factorization models; different techniques are compared with the selected evaluation criteria.