A Proposal for the Emotional Classification of an Album Using the Lyrics of its Songs

Abstract

Current volumes of data require automated methods for knowledge extraction. In particular, text mining deals with extracting knowledge from texts. In this paper we propose two algorithms to determine, from the lyrics of its songs, the emotional tendency of an album. We follow a hierarchical approach: the emotional categories to classify the albums group the emotional subcategories of the songs. This is reasonable because a song tends to be oriented to a specific emotional (sub) category. In this way, the emotional category of an album is a weighting of the emotional subcategories of its songs. This weighting can be customized by the musical analyst, which allows incorporating a subjective element in our proposal. Through a series of experiments we evaluated our algorithms. Although it is necessary to experiment with more data, our results show the expediency, consistency, and usefulness of our proposed algorithms.

Keywords

• Análisis musical, categorías emocionales, clasificación, minería de texto. results show the expediency, consistency, and usefulness of our proposed algorithms