Music Recommendation System Using Psychological Scale

Individual music preferences can generally depend on several factors, namely demographic data such as gender, age, etc., and more specific psychological factors like mood and personality. However, while mainstream music applications tend to consider the former, they usually ignore psychological factors that would otherwise allow for more accurate recommendations. In this paper, we attempt to develop a smart music recommendation system by comparing various machine-learning algorithms to recommend a genre based on the user’s current mood and personality traits, determined using a psychological scale, along with their basic demographic data. A custom dataset was built, where the TIPI scale was used to identify users’ personality traits, which we then fed into the models to generate recommendations. Based on the results, ensemble algorithms performed the best for the moods of “Happy”, “Gloomy”, and “Energetic", while semi-supervised algorithms achieved the best scores for the moods of “Stressed” and “Relaxed”.