Social networks with multiple relationship semantics

Social networks consist of a set of participants and the pairwise relationships between them. There are several different types of networks, such as directed networks, networks with typed edges, dynamic networks and signed networks, as well as any composition of different types of networks. We develop a novel way to analyze such networks by considering the qualitatively different social roles that each individual can play in a network. For example, in a directed network, the participants have two social roles - an incoming edge role and an outgoing edge role which are associated with the popularity and activity of each individual. Each role or status and the corresponding connections define a subgraph. We model the subgraph as a layer, and show how to weight the edges connecting the layers to produce a consistent spectral embedding. This embedding can be used to compute social network properties of graphs of different types, to predict edges, edge types, and edge direction, as well as to track the change of role over time. We illustrate the approaches using synthetic and real-world datasets.