Abstract
Recent years have seen a surge in interest in the investigation of various recommender systems that are based on social networks. Because users' preferences are likely to be similar to or influenced by those of their connected friends, the integration of the social relationships that already exist between users has the potential to improve the accuracy of the recommendation results. To increase the quality of suggestions, a collaborative filtering recommendation algorithm that incorporates the latent social trust model is developed. Global trust value and expert model in a social matrix, and improved Pearson coefficient model make up the new social trust model. Social matrix, trust propagation model, and improved Pearson coefficient are the primary elements that contribute to the overall value of global trust. It is essential to have a sparse rating matrix and even a sparser social matrix to uncover potential social trust linkages that can improve the quality of recommendations. Pearson coefficient takes into account the ratings that users have given different items, but it does not take into account the items that users have in common with one another. The experimental findings suggest that including the new model into the recommendation algorithm improves the recommendation effect.