Music recommendation using dynamic feedback and content-based filtering

The internet has become inundated with vast amounts of information, making it increasingly difficult for users to find precise and reliable content in their respective domains. Recommendation algorithms have emerged as a solution to this problem, enabling personalized content recommendations based on user preferences. One domain that greatly benefits from recommendation systems is music, with its diverse genres and countless songs. In this research paper, we present a novel approach to music recommendation that leverages user likes, plays, ratings, and other factors to generate personalized playlists. The proposed hybrid recommendation system analyses songs using six features and incorporates new features such as ELLT and DVS designed for dimensionality reduction to enhance accuracy. It continuously adapts to evolving user preferences and music trends, making personalised recommendations using cosine similarity and song weights. We test our approach through extensive experimentation and achieve promising precision, recall, and F1 scores. This contribution advances personalized content recommendation and hybrid recommendation systems, demonstrating their potential to improve music listening experiences.