Traditional recommendation systems typically rely solely on user content information to predict product ratings while ignoring the impact of other factors on recommendation effectiveness. Proposes a recommendation algorithm based on product attributes and social attributes that leverages product attribute information to obtain user preferences and product features, thereby reducing rating data sparsity. The algorithm also considers the effect of social attributes on the recommendation system and uses linear weighting with global rating average, item rating average, user rating average, and matrix decomposition to predict user ratings, thus improving prediction accuracy. The algorithm was tested on three public datasets, and the results indicate that it improves recommendation performance to a certain extent.