2011 IEEE International Conference on Multimedia and Expo

KOLMOGOROV COMPLEXITY IN LYRICS

Teppo Ahonen



Abstract

We examine how approximating Kolmogorov complexity in lyrics can be used for classification in music information retrieval. Using normalized compression distance as a similarity measure between lyrics, we perform various evaluations for large sets of lyrics spanning over several genres, lyricists and moods. Results suggest that straightforward compression-based similarity measuring can be applied to lyrics with competent performance.

Read Submission [124]