engleski

Detecting communities under constraints in directed acyclic networks

Community detection is one of the fundamental problems in complex networks theory with applications in many different branches of science. Many available algorithms for community detection in directed acyclic networks do not include analysis of the resulting set of communities, and those that do, mostly focus on factors like the number of communities and community stability, not on relations between communities. In this paper, we present an algorithm that, given the topological ordering of a directed acyclic network, produces an optimal division (in terms of modularity) for that ordering which allows the establishment of an ordering on the resulting set of communities. The algorithm is based on recursively placing of the vertices into appropriate communities, thus respecting the order of the vertices, and resulting in division with optimal modularity.

community detection ; directed acyclic network ; modularity ; topological ordering

nije evidentirano