Communicability geometry reveals antagonistic factions in signed networks

The study of signed networks has become increasingly popular over the last years. However, the field still has several open problems, like the definition of a well-defined distance metric, the detection of factions of allies, or the prediction of edge signs. In this work, we approach these questions through the notion of signed communicability. By taking into account all walks between two given nodes, we show that the communicability function can measure the effective level of alliance or conflict between them. Additionally, it induces an Euclidean distance that remains well-defined even in the presence of signed edges. Moreover, the communicability can be used to gain novel insights into the structural properties of empirical signed networks. In particular, we show how it can be applied to induce low-dimensional network embeddings, detect hidden factions, develop a hierarchy of alliances and infer the polarization of individual nodes.

Presential in the IFISC seminar room, Zoom stream at:

https://zoom.us/j/98286706234?pwd=bm1JUFVYcTJkaVl1VU55L0FiWDRIUT09