An Emotion-based Personalized Music Recommendation Framework for Emotion Improvement

Zhiyuan Liu, Wei Xu, Wenping Zhang*, Qiqi Jiang

*Corresponding author for this work

Research output: Contribution to journalJournal articleResearchpeer-review

Abstract

Music has a close relationship with people's emotion and mental status. Music recommendation has both economic and social benefits. Unfortunately, most existing music recommendation methods were constructed based on genre features (e.g., style and album), which cannot meet the emotional needs of listeners. Furthermore, the “filter bubble” effect may make the situation even worse, when a user seeks music for emotional support. In this study, we designed a novel emotion-based personalized music recommendation framework to meet users’ emotional needs and help improve their mental status. In our framework, we designed a LSTM-based model to select the most suitable music based on users’ mood in previous period and current emotion stimulus. A care factor was used to adjust the results so that users’ mental status could be improved by the recommendation. The empirical experiments and user study showed that the recommendations of our novel framework are precise and helpful for users.
Original languageEnglish
Article number103256
JournalInformation Processing & Management
Volume60
Issue number3
Number of pages12
ISSN0306-4573
DOIs
Publication statusPublished - May 2023

Bibliographical note

Published online: 23. December 2022

Keywords

  • Music emotion recognition
  • Emotional needs
  • Personalized music recommendation
  • Deep learning
  • User study

Cite this