Music recommendation system based on genre distance and user preference classification

초록

Background/Objectives: The personalized music recommendation services can support the user-favorite contents among various multimedia contents. In order to predict user-favorite songs, it is necessary to manage user preferences information and genre classification. Methods/Statistical analysis: We introduce the mechanism about the automatic management of the user preferences in the personalized music recommendation service. This system automatically extracts the user preference data from the user’s brain waves and audio features from music. Findings: In our study, a very short feature vector, obtained from low dimensional projection and already developed audio features, is used for music genre classification problem. We applied a distance metric learning algorithm in order to reduce the dimensionality of feature vector with a little performance degradation. Proposed user’s preference classifier achieved an overall accuracy of 81.07% in the binary preference classification for the KETI AFA2000 music corpus. Improvements/Applications: we could recognize the user’s satisfaction when we use brainwaves. This system can be applied to various audio devices, apps and services. © 2005 – ongoing JATIT & LLS

키워드