Consensus, polarization and clustering of opinions in social networks

Lin Li, Anna Scaglione, Ananthram Swami, Qing Zhao

Research output: Contribution to journalArticle

42 Citations (Scopus)

Abstract

We consider a variation of the Deffuant-Weisbuch model introduced by Deffuant et al. in 2000, to provide new analytical insights on the opinion dynamics in a social group. We model the trust that may exist between like-minded agents through a trust function, which is a discontinuous (hard-interaction) non-increasing function of the opinion distance. In this model, agents exchange their opinions with their neighbors and move their opinions closer to each other if they are like-minded (that is, the distance between opinions is smaller than a threshold). We first study the dynamics of opinion formation under random interactions with a fixed rate of communication between pairs of agents. Our goal is to analyze the convergence properties of the opinion dynamics and explore the underlying characteristics that mark the phase transition from opinion polarization to consensus. Furthermore, we extend the hard-interaction model to a strategic interaction model by considering a time-varying rate of interaction. In this model, social agents themselves decide the time and energy that should be expended on interacting each of their neighbors, based on their utility functions. The aim is to understand how and under what conditions clustering patterns emerge in opinion space. Extensive simulations are provided to validate the analytical results of both the hard-interaction model and the strategic interaction model. We also offer evidence that suggests the validity of the proposed model, using the location and monthly survey data collected in the Social Evolution experiment over a period of nine months.

Original languageEnglish (US)
Article number6517111
Pages (from-to)1072-1083
Number of pages12
JournalIEEE Journal on Selected Areas in Communications
Volume31
Issue number6
DOIs
StatePublished - 2013
Externally publishedYes

Fingerprint

Polarization
Phase transitions
Communication
Experiments

Keywords

  • consensus
  • continuous opinions
  • information aggregation
  • network learning
  • non-Bayesian models
  • opinion clusters
  • opinion diffusion
  • opinion dynamics
  • opinion formation
  • polarization
  • social networks

ASJC Scopus subject areas

  • Electrical and Electronic Engineering
  • Computer Networks and Communications

Cite this

Consensus, polarization and clustering of opinions in social networks. / Li, Lin; Scaglione, Anna; Swami, Ananthram; Zhao, Qing.

In: IEEE Journal on Selected Areas in Communications, Vol. 31, No. 6, 6517111, 2013, p. 1072-1083.

Research output: Contribution to journalArticle

@article{da0d7d39df2d4f8fa5df0b2e4b03e4aa,
title = "Consensus, polarization and clustering of opinions in social networks",
abstract = "We consider a variation of the Deffuant-Weisbuch model introduced by Deffuant et al. in 2000, to provide new analytical insights on the opinion dynamics in a social group. We model the trust that may exist between like-minded agents through a trust function, which is a discontinuous (hard-interaction) non-increasing function of the opinion distance. In this model, agents exchange their opinions with their neighbors and move their opinions closer to each other if they are like-minded (that is, the distance between opinions is smaller than a threshold). We first study the dynamics of opinion formation under random interactions with a fixed rate of communication between pairs of agents. Our goal is to analyze the convergence properties of the opinion dynamics and explore the underlying characteristics that mark the phase transition from opinion polarization to consensus. Furthermore, we extend the hard-interaction model to a strategic interaction model by considering a time-varying rate of interaction. In this model, social agents themselves decide the time and energy that should be expended on interacting each of their neighbors, based on their utility functions. The aim is to understand how and under what conditions clustering patterns emerge in opinion space. Extensive simulations are provided to validate the analytical results of both the hard-interaction model and the strategic interaction model. We also offer evidence that suggests the validity of the proposed model, using the location and monthly survey data collected in the Social Evolution experiment over a period of nine months.",
keywords = "consensus, continuous opinions, information aggregation, network learning, non-Bayesian models, opinion clusters, opinion diffusion, opinion dynamics, opinion formation, polarization, social networks",
author = "Lin Li and Anna Scaglione and Ananthram Swami and Qing Zhao",
year = "2013",
doi = "10.1109/JSAC.2013.130609",
language = "English (US)",
volume = "31",
pages = "1072--1083",
journal = "IEEE Journal on Selected Areas in Communications",
issn = "0733-8716",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "6",

}

TY - JOUR

T1 - Consensus, polarization and clustering of opinions in social networks

AU - Li, Lin

AU - Scaglione, Anna

AU - Swami, Ananthram

AU - Zhao, Qing

PY - 2013

Y1 - 2013

N2 - We consider a variation of the Deffuant-Weisbuch model introduced by Deffuant et al. in 2000, to provide new analytical insights on the opinion dynamics in a social group. We model the trust that may exist between like-minded agents through a trust function, which is a discontinuous (hard-interaction) non-increasing function of the opinion distance. In this model, agents exchange their opinions with their neighbors and move their opinions closer to each other if they are like-minded (that is, the distance between opinions is smaller than a threshold). We first study the dynamics of opinion formation under random interactions with a fixed rate of communication between pairs of agents. Our goal is to analyze the convergence properties of the opinion dynamics and explore the underlying characteristics that mark the phase transition from opinion polarization to consensus. Furthermore, we extend the hard-interaction model to a strategic interaction model by considering a time-varying rate of interaction. In this model, social agents themselves decide the time and energy that should be expended on interacting each of their neighbors, based on their utility functions. The aim is to understand how and under what conditions clustering patterns emerge in opinion space. Extensive simulations are provided to validate the analytical results of both the hard-interaction model and the strategic interaction model. We also offer evidence that suggests the validity of the proposed model, using the location and monthly survey data collected in the Social Evolution experiment over a period of nine months.

AB - We consider a variation of the Deffuant-Weisbuch model introduced by Deffuant et al. in 2000, to provide new analytical insights on the opinion dynamics in a social group. We model the trust that may exist between like-minded agents through a trust function, which is a discontinuous (hard-interaction) non-increasing function of the opinion distance. In this model, agents exchange their opinions with their neighbors and move their opinions closer to each other if they are like-minded (that is, the distance between opinions is smaller than a threshold). We first study the dynamics of opinion formation under random interactions with a fixed rate of communication between pairs of agents. Our goal is to analyze the convergence properties of the opinion dynamics and explore the underlying characteristics that mark the phase transition from opinion polarization to consensus. Furthermore, we extend the hard-interaction model to a strategic interaction model by considering a time-varying rate of interaction. In this model, social agents themselves decide the time and energy that should be expended on interacting each of their neighbors, based on their utility functions. The aim is to understand how and under what conditions clustering patterns emerge in opinion space. Extensive simulations are provided to validate the analytical results of both the hard-interaction model and the strategic interaction model. We also offer evidence that suggests the validity of the proposed model, using the location and monthly survey data collected in the Social Evolution experiment over a period of nine months.

KW - consensus

KW - continuous opinions

KW - information aggregation

KW - network learning

KW - non-Bayesian models

KW - opinion clusters

KW - opinion diffusion

KW - opinion dynamics

KW - opinion formation

KW - polarization

KW - social networks

UR - http://www.scopus.com/inward/record.url?scp=84878118133&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84878118133&partnerID=8YFLogxK

U2 - 10.1109/JSAC.2013.130609

DO - 10.1109/JSAC.2013.130609

M3 - Article

AN - SCOPUS:84878118133

VL - 31

SP - 1072

EP - 1083

JO - IEEE Journal on Selected Areas in Communications

JF - IEEE Journal on Selected Areas in Communications

SN - 0733-8716

IS - 6

M1 - 6517111

ER -