Polarization in cooperative networks through optimal placement of informed agents

Abstract

The subject of opinion dynamics is one of the challenging topics in various fields of social networks, economics, politics, etc. Various models have been introduced to model how individuals can influence each other. Specially, the behavior of heterogeneous individuals such as stubborn or informed agents has been deeply investigated. Moreover, it has been shown that consensus on an arbitrary desired opinion can be formed with the help of informed agents. In this paper, polarization, fragmentation, and diversity of opinion are considered. The polarization of a given society increases the motivation of members for involvement; while the diversity of opinions is necessary to have a wise crowd. Fragmentation can be considered as a middle goal to reach the diversity of opinions. In order to reach the mentioned goals, informed agents should be carefully chosen. Therefore, a mathematical optimization technique is developed to analytically solve the problem of the informed agents selection. As a result, we face with a zero norm minimization problem which is NP-hard. Two methods are utilized to handle this problem. These methods are evaluated on a sample network in order to reach the mentioned goals. Finally, we investigate how the selected informed agents can lead the society toward the desired opinion. It is shown that confidence level has a meaningful effect on the success rate of informed agents.