A survey on the fusion process in opinion dynamics

Highlights

• We introduce the basic framework of fusion process in opinion dynamics.

• We review the basic models, extensions and applications in opinion dynamics.

• We propose some open problems regarding the fusion in opinion dynamics.

Abstract

Opinion dynamics is a fusion process of individual opinions, in which a group of interacting agents continuously fuse their opinions on the same issue based on established fusion rules to reach a consensus, polarization, or fragmentation in the final stage. To date, many studies have been conducted on opinion dynamics. To provide a clear perspective on the fusion process in opinion dynamics, this paper presents a review of the framework and formulation of opinion dynamics as well as some basic models, extensions, and applications. Based on the insights gained from prior studies, several open problems are proposed for future research.

Introduction

In social phenomena, humans are the basic elements, and human behaviors depend on many variables. The most important factors behind human behavior are opinions and beliefs that drive actions [1]. Therefore, understanding the process of opinion fusion is key to explaining human choices.

Opinion dynamics is the study of the opinion fusion process [2] through interactions among a group of agents. Opinion dynamics research originated in France [3], and some interesting opinion dynamics models with different opinion formats and fusion rules have since been proposed, such as the DeGroot model [4], [5], voter model [6], [7], [8], [9], Sznajd model [10], [11], majority rule model [12], [13], [14], Friedkin and Johnsen model [15], [16], bounded confidence model [17], [18], [19], and continuous opinions and discrete actions model [20], [21].

Opinion dynamics models are usually composed of a few basic elements - opinion expression formats, fusion rules, and opinion dynamics environments - and focus on three varieties of stabilized patterns: consensus, polarization, and fragmentation [18]. In the existing research, according to the different opinion formats expressed by agents, the models of opinion dynamics can be divided into two types: continuous opinion models (e.g., [4], [5], [17], [18]) and discrete opinion models (e.g., [6], [7], [8], [10], [12], [13], [22]). Moreover, an agent will neither simply share nor completely disregard the opinions of other agents but will take these opinions into account to a certain extent in forming his/her new opinions in a process defined by a fusion rule. The fusion process in opinion dynamics is influenced by different opinion dynamics environments (e.g., social networks [23], [24], [25] and noise [26], [27], etc.).

To provide a clear perspective on the fusion process in opinion dynamics, this paper presents a review of opinion dynamics. Moreover, with respect to insights gained from previous research, we aim to identify open problems and new directions for future research.

The rest of this paper is organized as follows. In Section 2, we introduce the framework and formulation of the fusion process in opinion dynamics. In Section 3, we review some basic models in opinion dynamics. Next, in Section 4, we introduce some extensions of opinion dynamics models. Following this, we survey the applications of opinions dynamics in Section 5. Subsequently, we provide and analyze open problems and new directions in opinion dynamics in Section 6. Finally, we present our conclusions in Section 7.