DESIGN OF RECOMMENDER SYSTEMS IN E-COMMERCE

Abstract

This research presents the design of a recommender system for an online store. The system utilizes user behavior data and purchase history to generate personalized product suggestions. Its main components include a recommender server and client modules. The recommender server performs the core recommendation logic independently of the online store, while the client modules integrate with the store’s interface to display recommendations to users. For data analysis and recommendation generation, the system employs a combination of K-Means clustering and collaborative filtering. K-Means clustering groups users based on similarities in their purchase history, while collaborative filtering suggests products based on the purchases of other users within the same cluster. The paper also addresses implementation challenges encountered during system development, including the selection of appropriate libraries and performance optimization. The results indicate that the system provides effective product recommendations and shows strong potential for future enhancements, such as the incorporation of additional user data and exploration of more advanced recommendation algorithms. Future work may focus on algorithm optimization, dataset expansion, and experimentation with emerging technologies.