A Multilayer General Type-2 Fuzzy Community Detection Model in Large-Scale Social Networks
Naderipour Mansoureh, Fazel Zarandi Mohammad Hossein, Susan Bastani
Abstract
In this article, a novel general type-2 fuzzy community detection model is presented based on different interactions in multilayer graphs to detect communities in a graph with an interaction between nodes, called a multiplex graph considering both structural and attribute similarities. The focus of this model is mostly on the uncertainty related to the interactions between nodes. This model regards a set of layers as the variety of relations among a set of nodes. Based on multilayers, the membership functions of the communities are delineated as “general type-2 fuzzy sets” comprising “primary and secondary variables.” The former indicate the degree of belonging to communities in the multiplex graph based on different interactions in multilayers, and the latter indicate the degree of belonging of the multilayers to the multiplex graph. In the suggested community detection algorithm, no type-reduction or defuzzification exists to update the community prototypes. Moreover, an index, in accordance with the suggested model, is defined to validate the suggested model and its results; and several experimental results on different network sizes including large-scale social networks are given to determine the efficiency of the suggested model compared to the other models.