Earticle

초록

영어

The digital music market has been growing significantly in the past years. In music streaming services, music recommendation plays a key role, but Korean users’ recognition about their music service is not high because the service’s recommendation accuracy is not good. Therefore, this paper suggests technique to predict the user’s musical taste. This technique proceeds through a four-step process; data collection, data pre-processing, feature extraction, and machine learning. Collection of data was taken from TOP 100 chart in ‘Melon’, the number one music service provider in Korea from December 2013 to March 2015. Then, collected MP3 file format is converted into WAV file format. In the stage of feature extraction, we classify the genre from the music’s metadata and extract factors that can be taken using STFT’s ZCR, Spectral Rolloff, Spectral Flux. In the stage of machine learning, we produce a prediction model in a variety of classification techniques. To measure the performance of the created prediction model, 456 data were used for training dataset and 130 data were used for validation dataset. Since the results of experiment show an average of 78% of accuracy, the proposed technique seems to be effective.

목차

Abstract
 1. Introduction
 2. Related Research
  2.1 Related Researches about Music Recommendation
  2.2 Extracting Features of Music
 3. Suggested Technique
  3.1. Data Collection
  3.2. Data Preprocessing
  3.3. Music Feature Extraction
  3.4. Machine Learning
 4. Experimental Results and Analysis
  4.1. Result Analysis and Conclusion
 References

키워드

저자

• Minseo Gong [ Graduate School of Software, Soongsil University, Seoul, Korea ]
• Jae-Yoon Cheon [ Graduate School of IT Policy & Management, Soongsil University, Seoul, Korea ]
• Young-Suk Park [ Graduate School of IT Policy & Management, Soongsil University, Seoul, Korea ]
• Jeawon Park [ Graduate School of Software, Soongsil University, Seoul, Korea ]
• Jaehyun Choi [ Graduate School of Software, Soongsil University, Seoul, Korea ] Corresponding author

참고문헌

발행기관

간행물

표지보기 관심저널 등록

• 간행물명

  International Journal of Multimedia and Ubiquitous Engineering

• 간기

  월간

• pISSN

  1975-0080

• 수록기간

  2008~2016

• 등재여부

  SCOPUS

• 십진분류

  KDC 505 DDC 605