Visual Analysis of Bipartite Networks

Since a heterogeneous network in general is too complex to be visually analyzed as graphs, we propose the alternative method for visualizing a heterogeneous network as a non-graph visualization. In this paper, we study the bipartite network which is one of the most common heterogeneous networks. Visualization of a few high-degree vertices in a bipartite network will make the graph more complex to be visualized. We present the visual analysis of bipartite networks by a novel method, called LBI distribution. We have modified from a previous technique which represents summarized networks by their features. Some features were modified in order to be compatible with bipartite networks by analyzing the properties and feature structures. Then we visualize those features on the LBI (1,1,1) plane. From the results of this visualization, we can visually analyze by describing characteristics of a bipartite network as a whole. Furthermore, we can detect outliers, such as high-degree vertices.