The Compound Graph: A Case Study for Community Visualisation in Social Networks

Abstract

This paper builds on previous work which aimed at providing a graph-based visual exploration of melodic relationships (tune families) within collections of traditional music. Here, using a community detection algorithm, potential tune families can be readily identified. However, the richer the information contained in the graph, the more difficult it is for the visualisation algorithms to operate successfully. Therefore, an approach is proposed which uses modified versions of the graph both to enhance the community detection results and, more importantly, restructure the graph, by creating a compound graph, to reveal the communities visually. Finally, the wider applicability of the technique is considered.