Design and development of emotion-based user-customized music classification system

Abstract

The emotion engineering is a field of science intending to scientifically interpret emotions of humans and to design and develop products and environment system that are appropriate on them. Hereupon, it is a current trend that a study dealing with removal of stress of humans is being researched with emotion engineering. The objective of this study is to research the user customized music recommending system by utilizing emotion engineering. Especially, online music service companies in Korea are currently providing a user customized emotion-based searching function. This is a service system making the opinion recorded after a user listens to the music as DB, classifying the emotion, and making a recommendation. This study has designed and developed a system that automatically supported the customized service through user customized emotion classification on popular music in Korea. Implementation method of suggested system hereof is to extract characteristics of the music classifying the code information and BPM, mapping two-dimensional emotion model by Theyer, and extracting emotional adjectives. For the experiment and examination, total 180 songs were selected in each type of emotionfrom digital music proceeding comparison analysis and experiment on the emotionclassification system suggested in this study. As a result of the experiment, emotional adjectives of being 'sad', 'excited' and 'pleasure' frequently felt by users represented 80% of similarity when comparing with the system and investigation resources suggested in this study. However, adjectives such as 'surprising,' angry' and 'mysterious' represented a low similarity of 10%. As for follow-up studies, it is planned to proceed a research about further development of system suggested by current study and classification of emotions from humans by utilizing various types of biosignals.