Hybrid Deep Learning Model for Mitigating Cold Start and Data Sparsity Challenges in Personalized News Recommendation Systems
Abstract
The growing volume of online news content presents a significant challenge in delivering personalized and contextually relevant information to users. Traditional recommendation systems, primarily based on collaborative and content-based filtering, often suffer from cold start and data sparsity issues that degrade recommendation quality and user satisfaction. This study proposes a hybrid deep learning model that integrates collaborative filtering, content-based filtering, and neural network architectures to enhance the accuracy and adaptability of personalized news recommendations. The hybrid framework leverages user behavior data and semantic feature extraction from news articles to generate richer representations that effectively capture user–item relationships. Experimental evaluations using real-world datasets demonstrate that the proposed model significantly outperforms traditional approaches in terms of precision, recall, and personalization metrics. The results highlight the potential of hybrid deep learning frameworks to overcome inherent limitations in conventional recommender systems, thereby contributing to improved user engagement and intelligent content delivery in digital media environments.