부산대학교 도서관

E-Commerce is a prominent application of information technology in business, offering convenience in transactions. However, as more products and users join the platform, the complexity of E-Commerce systems increases. This necessitates the implementation of a recommendation system to enhance the user experience and address individual preferences. E-Commerce platforms often face the challenge of the cold start problem, where new products are introduced to the platform or when there is limited information available about new users. In such cases, traditional recommendation systems struggle to provide accurate recommendations due to the lack of historical data or user preferences. In this research, a comprehensive framework for an efficient multi-mode Hybrid Recommendation System is proposed, focusing on addressing the cold start problem. The system leverages user behavior tracking, specifically search history and product visit history, to capture user preferences effectively. Through experimental evaluation, the system demonstrates its adaptability to different dataset conditions, achieving a high precision rate of 90%. This research addresses the challenges of data sparsity and the cold start problem in recommendation systems, particularly in the E-Commerce context. The flexibility of the recommendation system ensures optimal recommendations in various scenarios, with potential applications beyond E-Commerce.