DEVELOPMENT OF MUSIC RECOMMENDATION SYSTEM WITH HYBRID FILTRATION APPROACH

• Main Author: Kennedy, Ricky
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/50184
• Institution: Institut Teknologi Bandung

Music recommender system which using a hybrid filtration approach combines from content-based and collaborative filtering learning algorithms. Hybrid filtration gives better results than only using content-based or collaborative filtering. Therefore, this final project was developing a music recommender system with a hybrid approach. In this work, the content-based approach is simplified than the previous research. However, they are expected to provide appropriate recommendations for users. for the content-based approach only using social genes from the music. To find similarity between song and user on content-based and collaborative filtering used KNN algorithm. Results from both approaches merge become the learning result of the hybrid filtration approach. Based on the research that has been done, music recommendation with hybrid filtration gives a good result. Respondents give a score of 85.5 for relevance music to the user, give a score of 89.4 to give a new song to the user. With both scores, the recommendation system that was built was able to meet the objectives of the recommendation system. However, there are still shortcomings that the music recommendation system has built which is the lack of diversity of songs due to the limited dataset used.