Aggregating Contextual Information for Multi-Criteria Online Music Recommendations

This paper introduces CAMCMusic, a novel context-aware multi-criteria music recommendation system designed to address these limitations without relying on user-specific attributes, music features, or explicit user ratings. CAMCMusic integrates contextual information into a multi-criteria decision-making framework to align music genre recommendations with the user’s contextual situation for generating Top-N music recommendations. CAMCMusic begins by assessing the relevance of contextual information based on the relationship between three key elements: the user, the music genre, and the user’s context. Subsequently, we employ an aggregation technique to reveal the connection between context and overall ratings. On the MusiClef dataset, CAMCMusic achieves a precision@5 of 0.61, a recall@5 of 0.77, and F1@5 of 0.68, surpassing the next best performer, SVD++, by 52.5%, 11.6%, and 36%, respectively. On the CAL500 dataset, CAMCMusic outperforms the FM algorithm by 11.9% in precision, 16.4% in recall, and 20% in F1-Score. CAMCMusic significantly enhances recommendation outcomes by effectively capturing and utilizing the relationships between user contexts and music genres, thus providing a robust solution to the cold start problem while maintaining high recommendation quality and user satisfaction. The findings underscore the potential of combining context-awareness with multi-criteria decision-making (MCDM) to advance the state of the art in music recommendation systems.