The rise of social networks has spurred users to share their experiences, influencing recommender systems. This paper integrates personal and interpersonal factors into a unified recommendation model using probabilistic matrix factorization. Personal interest tailors recommendations for experienced users, while interpersonal interest similarity and influence aid cold start users. Experiments on Yelp, Movie Lens, and Douban Movie datasets demonstrate superior performance. Keywords: Interpersonal influence, personal interest, recommender system, social networks.

• Probabilistic Matrix Factorization
• Recommender Systems
• Social Networks