Extending finite mixtures of latent trait analyzers for bipartite networks

The paper extends the Mixture of Latent Trait Analyzers (MLTA) for clus tering bipartite networks to account for nodal attributes. Bipartite networks are par ticularly useful to represent relations between disjoint sets of nodes, called sending and receiving nodes. The MLTA model is able not only to cluster the sending nodes of a bipartite network, but also capture the latent variability of network connections within each group. We extend this approach by including nodal attributes to study how nodes’ characteristics affect the group membership probability. A simulation study is conducted to evaluate the proposed approach.