An algorithm with tunable resolution for detecting overlapping communities in bipartite networks

We propose an extension of a recently introduced algorithm for detecting overlapping communities in bipartite networks based on the maximization of the Fuzzy Entropic Barber Modularity. Similar to Fuzzy clustering, nodes are given a membership distribution which represents how much a node fits into each of the communities. Together with the fuzzyfied version of the Barber modularity, in a linear combination the Fuzzy Entropic Barber Modularity optimizes an entropic term which accounts for the uncertainty of the bipartite network itself, and also a new term here introduced to control the number of communities. The corresponding temperature and resolution parameters allow for an online optimization of the number of communities and fuzzy memberships.