Overcoming the Cold Start Problem of Collaborative Filtering Recommendation Systems: A Hybrid Model Combining Online Clustering and Sequential Learning

Recommendation systems are widely used in various platforms to mitigate information overload by providing personalized content. However, they often suffer from the cold-start problem, in which a lack of historical data for new users or items limits recommendation accuracy. This paper proposes a hybrid recommendation framework that integrates online clustering and sequential learning to address this challenge. The clustering model groups users based on demographics and proportional preferences, while a neural network captures behavioral patterns from sequential interactions. The outputs are fused using cosine similarity between the user preference vector and item category vectors, generating dynamic recommendations. Experiments conducted on the MovieLens <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{1 M}$</tex> dataset under a simulated cold-start scenario demonstrate that the proposed framework achieves higher precision (0.74) and accuracy (0.82) compared to the selected baselines. These results highlight the model's effectiveness in mitigating the cold start problem. Future work will explore advanced clustering techniques and deep learning-based sequential models to further enhance recommendation quality.