engleski

Music Lyrics Mining and Sentiment Analysis

Sentiment analysis of lyrics is an important part of music recommendation systems as it is used for classification od songs. However, finding accurate classifiers remains a challenging task due to lack of comprehensive lyrics datasets. Based on a large dataset of lyrics, this thesis explores the process of automatic pre-processing of datasets and sentiment analysis based on use of affective lexicons. The results indicate what kind of affective dimensions can be recognized, and how they can be interpreted.

data mining ; sentiment analysis ; music lyrics ; affective lexicons ; visualization

nije evidentirano