Current Trends in Recommender Systems: A Survey of Approaches and Future Directions

Creative Commons License

Akkaya B.

Journal of Computer Science, cilt.10, sa.1, ss.53-91, 2025 (Hakemli Dergi)

Özet

This paper discusses the growing importance of recommender systems in enhancing user experience and information access in digital environments. It identifies challenges such as data sparsity, the cold-start problem, and scalability, emphasizing the need for advanced machine learning techniques. Various methodologies are explored, including collaborative filtering, content-based filtering, and hybrid approaches. Innovations like graph-based collaborative filtering, graph neural networks, and deep learning are highlighted for addressing data sparsity and complex data relationships. The paper also emphasizes attention mechanisms and sequential modeling to resolve the cold-start problem and adapt to changing user preferences. It stresses the significance of explainable AI for building user trust and transparency. Looking ahead, the paper anticipates advancements in cross-domain recommendation models and the integration of diverse data sources to enhance personalization and relevance. Overall, it advocates for sophisticated methodologies to overcome challenges and improve user satisfaction in digital platforms, underscoring the role of innovation in the future of recommendation technologies.