The design of a mood-driven chinese song recommendation system: Combining valence-based and polarity-based sentiment analysis on lyrics
Document Type
Conference Proceeding
Publication Date
1-1-2018
Abstract
Recommendation system (RS) can be lucrative in attracting and retaining users. A more intelligent RS would further make recommendations based on the user’s current mood. Driven by previous research that emotion is one of the dominating factors behind decision making, we present in this paper, an emotion-aware Chinese music RS which takes a user’s emotion state as an input to filter recommendation list accordingly. In particular, the system asks the users for a sentence describing their mood state then the system will try to recommend songs with the similar mood to the users. Such system can be used when lack of user data can be integrated into the music application to improve the overall user experience. Our system differentiates itself from most of the previous ones. In that the computational cost has been significantly reduced, thanks to the lightweight design of the core recommendation technique.
Publication Title
Advances in Intelligent Systems and Computing
First Page Number
669
Last Page Number
678
DOI
10.1007/978-3-030-01057-7_51
Recommended Citation
Sun, Qiming and Tang, Tiffany Y., "The design of a mood-driven chinese song recommendation system: Combining valence-based and polarity-based sentiment analysis on lyrics" (2018). Kean Publications. 1536.
https://digitalcommons.kean.edu/keanpublications/1536