On efficient use of entropy centrality for social network analysis and community detection

• We motivate and interpret entropy centrality metric for path and walk based flows in networks.
• We showcase the flexibility of the metric in allowing varying locality analyses.
• We present an accurate community detection algorithm supported by experimental tests.

This paper motivates and interprets entropy centrality, the measure understood as the entropy of flow destination in a network. The paper defines a variation of this measure based on a discrete, random Markovian transfer process and showcases its increased utility over the originally introduced path-based network entropy centrality. The re-defined entropy centrality allows for varying locality in centrality analyses, thereby distinguishing locally central and globally central network nodes. It also leads to a flexible and efficient iterative community detection method. Computational experiments for clustering problems with known ground truth showcase the effectiveness of the presented approach.