Using Network Science to Explore the Organization of Patterns Within Large Music Corpora

Network science can be used to explore and quantify organizational structures within large collections of similar units. Here I demonstrate how to convert music corpora into networks where those units represent short interval patterns. As an example of network structure analysis, I compared patterns within the Weimar Jazz Database to a version of the Essen Folk Song corpus. The analysis showed that both corpora contain mostly patterns that are sparsely organized around much fewer highly connected hubs. This organizational principle has been termed “scale-free” and has been used to describe other types of naturally occurring networks. I further explore potential network growth models previously used in language research to describe how children build vocabulary. Finally, I suggest other types of music related networks based on different connectional principles. Applying network analysis to music corpus research may reveal organizational principles not previously explored.