Enhancing E-Commerce Recommender Systems by Integrating Content-Based Filtering, Collaborative Filtering, and Neural Network Techniques

Recommender systems are recognized as essential tools for enhancing user experience and increasing engagement in e-commerce platforms. However, traditional methods, such as content-based filtering and collaborative filtering, face challenges like the cold-start problem and limited diversity in recommendations. This research presents a hybrid recommender system that leverages two distinct approaches to improve recommendation accuracy. The system employs textual analysis of products using the TF-IDF method and cosine similarity to identify similar products, and these results are combined with collaborative filtering, which analyzes user interactions. To integrate the results, a neural network model dynamically calculates final scores for hybrid recommendations. The proposed system was evaluated using real-world data from an online retail dataset, demonstrating that the hybrid model outperforms standalone methods. In addition to improving the accuracy and quality of recommendations, the system provides more diverse and relevant suggestions, leading to increased customer satisfaction and enhanced marketing strategies. The findings of this study emphasize the advantages of hybrid systems and their potential to optimize sales and enhance the online shopping experience.

Enhancing E-Commerce Recommender Systems by Integrating Content-Based Filtering, Collaborative Filtering, and Neural Network Techniques Keywords:

Recommender systems , Content-based filtering , Collaborative filtering , Hybrid system , TF-IDF , Cosine similarity , Neural network , E-commerce