One of the challenges faced by today's societies is to deal with the growing polarization. Here, we propose an agent-based model incorporating theories of structural balance and homophily. Most of the literature identify a structurally balanced state as a polarized one. However, we show that these two states are not always equivalent. We study a multilayer system with one layer related to the agents' relations and the others to the similarity between agents. We define the polarization as the state with two or more enemy groups in the relation layer and we study the influence of different types of homophily in the similarity layers and homophily strength on the fates of the system. We identify homophily types that are most efficient in limiting the system polarization.
This research received funding from National Science Centre, Poland Grant No. 2019/01/Y/ST2/00058.