Learning Automata for Complex Social Networks

Abstract

Since learning automata are suitable techniques for modelling, learning, controlling and solving dynamic and distributed problems in unknown environments, in this chapter, we present a brief description of some learning automata based algorithms for applications of social network analysis. In this regards, first we introduce complex social networks and stochastic graphs as a graph model of social networks for analysis objectives. Then, several algorithms using learning automata are introduced for solving stochastic graph problems (i.e., maximum clique problem and minimum vertex covering), computing an estimation of network centrality measures, sampling from networks and community detection. In LA-based algorithms for social networks when stochastic graphs are used as a graph model, learning automata are used by the network as a means for observing the time varying parameters of the network for the purpose of network analysis. In complex social networks, the aim of the LA-based algorithms is to collect information from the social network in order to find good estimates for the network parameters and measurements using as few numbers of samples as possible.