Advances in Network Clustering and Blockmodeling

9Partitioning Multimode Networks

Martin G Everett¹, and Stephen P Borgatti²

¹University of Manchester

²University of Kentucky

9.1 Introduction

Most networks examined so far involve connections between nodes all of the same type, known as one-mode networks. There are circumstances in which the nodes are of different types and the connections are only between different types of nodes, and not between nodes of the same type. We refer to these as multimode networks, some authors call these multiway networks. A simple example consists of nodes made up of authors and journals. An author is connected to a journal if they have published a paper in that journal. Since we have two types of nodes, authors and journals, this results in a two-mode network. There are many examples of two-mode networks, such as people attending events [16], legislators being members of committees [38], directors serving on boards [13], companies collaborating on projects [39] etc. In principle there is no reason to limit the node types to two, we could have three or more. An example of a three mode dataset would be criminal by crime by victim. Such datasets are less common and we shall concentrate at first on two mode datasets and discuss general multimode approaches later in the chapter.

For clarity of exposition, when considering two-mode data we shall refer to one mode as actors and the other mode as events. The resultant network will form a bipartite network, that is a network in which the nodes can ...