DEVELOPING AN E-COMMERCE PLATFORM WITH RECOMMENDATION ENGINE

Abstract

E-commerce platforms have become an essential component of modern digital business, providing users with convenient access to products and enabling companies to optimize online sales. However, as the volume of products and user interactions continues to grow, traditional rule-based recommendation methods are no longer sufficient to deliver personalized experiences. This thesis presents the design and development of an e-commerce platform integrated with a hybrid recommendation engine that enhances product discovery, improves user engagement, and increases overall system efficiency. The proposed system combines both content-based filtering and collaborative filtering techniques, supported by additional heuristic signals such as popularity scores and user behavioral patterns. A dedicated dataprocessing pipeline records user interactions—including product views, likes, addto-cart actions, and purchases—to construct detailed user profiles and item feature representations. Machine learning models, including Alternating Least Squares (ALS) and LightFM, are applied to generate personalized recommendations and learn latent relationships between users and products. The platform is implemented using modern web technologies such as Next.js for the frontend, Supabase for data management and authentication, and PostgreSQL functions to support real-time analytics. A modular microservice-style architecture ensures system scalability and allows the recommendation engine to update continuously as new data is collected. Experimental evaluations demonstrate that the hybrid approach significantly improves recommendation quality compared to baseline rule-based methods, particularly in scenarios with sparse interaction data. This study highlights the importance of integrating machine learning into ecommerce systems and provides a practical framework for building scalable, datadriven recommendation features. The results show that a thoughtfully designed hybrid recommendation engine can enhance user satisfaction and support business growth by delivering more relevant and personalized shopping experiences.