An Efficient Collaborative Filtering with Repurchase Probabilistic Model for E-Commerce Recommendation

Recommender systems have been extensively researched and implemented across various domains to address information overload. While substantial research exists on recommender systems for movies and e-commerce, there has been relatively less focus on recommender systems for virtual products offered by telecommunication operators. In light of user privacy protection and cross-platform isolation, we propose a recommendation system solution that relies solely on order record data corresponding to the Rights Platform. Initially, a user repurchase model is developed using big data technique to represent user interest preferences, referred to as privacy feedback rating. This model is then integrated with collaborative filtering algorithm to deliver personalized recommendation services to users. Furthermore, we experimented with our proposed method on a real-world dataset and current online platform for virtual products purchase. The experimental results confirmed the effectiveness of the method, demonstrating strong purchase conversion rates across various scenarios.