The authors propose an online social trust model that consists of public trust users and private trust users for use in a recommender algorithm. The public trust users are defined as users who have a large number of fans in the social network. Their item recommendations are more likely accepted and trusted by the public. The private trust users are defined as a set of online friends in the social network. The authors construct a Bayesian network (BN) to model the trust relationship among these public and private users. The trust degree is the reliability of an item recommendation, represented as a conditional probability. With inference on the BN, they derive further trust relationships. Their recommendations are based on the similarity between the trust users and the target user in terms of items or reviews and on the proximity of time of accesses by two users. The items accessed in closer time proximity may be of more interest to the target user. This social trust-based recommendation method is shown to be more effective for Yelp data, compared to a few alternative methods.

The time proximity needs more justification through social theoretical evidence. The interpretation of the temporal proximity results shows that the more data is used, the lower the mean errors. It is difficult to see whether the time proximity supports the better recommendation results. The paper also suffers from a lack of concrete examples, various typographical errors, and incorrect references to figures.