Professional Documents
Culture Documents
2016.proceedings Cmmse 1
2016.proceedings Cmmse 1
Abstract
1 Introduction
The specialized literature has already proposed various mathematical models to investigate
the dynamics of opinions in social networks, some of the most popular being the models
proposed by Sznajd-Weron and Sznajd [8], the voter model in complex networks [7], the
CMMSE
c 866 ISBN: 978-84-608-6082-2
Consensus formation
models investigated by Deffuant et al. [1] and the Hegselmann–Krause system [3]. In all of
these models, the opinion of each agent is a consequence of interactions between members
of the social network (for example, such interactions are considered in the weights matrix of
the De Groot model [2], and Hegselmann–Krause system). Moreover, the determination of
conditions that give rise to consensus has been a common topic of many research articles,
including those cited above.
It is important to point out that the notion of consensus in sociology has been compared
in relevance to the concept of energy in physics [6], and the investigation of the existence
of consensus has been carried out on discrete spaces, voter models and bounded confidence
systems alike. However, when personal values and beliefs are directly involved in an opinion,
the postures are often defensive and the opinions steady, and many of the classical models
become inapplicable in such circumstances. In view of those limitations, the purpose of
this work is to investigate the formation of consensus in a new model that describes the
dynamics of opinions on controversial topics.
Opinion dynamics has been frequently modeled using fully discrete dynamical systems
[4]. Indeed, realistic settings actually consider the presence of a finite number of agents as
well as discrete interactions within finite subgroups of the social network at discrete instants
of time, whence the use of systems of coupled difference equations is justified in the practice.
In the present study, we will propose a system of difference equations to model the dynamics
of opinions on controversial subjects. Our model may conveniently incorporate attractors
and repellers in order to describe strong postures of the agents, together with an individual
parameter proposed previously in [9]. The inclusion of that additional parameter gives the
advantage of mimicking the existence of strong personal attitudes and values often motivated
by hard changing cultural paradigms in each of the agents of the network. In that work,
the author justified its use by social and/or environmental reasons, but we will employ
it in the present manuscript to prescribe the cultural factors as well as personal values,
knowledge, predisposition, etc. In summary, the parameter used in [9] will be employed
here to represent strong attitudes acquired during the lifetime of each individual, and we
will denominate it the cultural baggage or, simply, the baggage of the agent.
2 Preliminaries
2.1 Nomenclature
Throughout this manuscript, we will suppose that the temporal interval [0, ∞) is partitioned
into a (not necessarily uniform) sequence of points 0 = t0 < t1 . . . < tn < . . . , for n ∈
Z+ ∪ {0}. We will assume that X is a connected subset of R that represents the space of
opinions. Meanwhile, the set K ⊆ R will represent the collection of possible values of the
cultural baggages of the agents, and it will be called the space of baggages. In this work,
we will fix an even number N of agents conforming the population under investigation,
CMMSE
c 867 ISBN: 978-84-608-6082-2
Medina et. al.
and let ΩN = {1, 2, . . . , N }. At the nth temporal step, the ith agent will be completely
characterized by an ordered pair xin = (xin , κi ) consisting of an opinion xin ∈ X and a
baggage κi ∈ K, for each i ∈ ΩN .
Following the approach of [2], the social network under consideration will be described
by the N × N square matrix of absolute weights, cij .
For each pair of indexes i, j ∈ ΩN , the number cij will represent the trust that the ith
agent has on the opinion of the jth agent. The weights do not vary in time, and must satisfy
the condition N j=1 cij = 1, for each i ∈ ΩN . Moreover, we will assume that the inequalities
0 ≤ cij < 1 and 0 < cii ≤ 1 hold for any pair of different indexes i, j ∈ ΩN .
Throughout, we let ǫ be a positive number which will be called the coefficient of affinity
or the confidence level [3], and it will be used to restrict the interaction between the agents
of the population. We say that the opinions of the ith and the jth agents at time tn are
affine if |xin − xjn | < ǫ, for each i, j ∈ ΩN and n ∈ Z+ ∪ {0}. Equivalently, we say simply
that xin and xjn are affine.
Let Ξ : X × K → R used to model a process of internal reflection for the individual
opinion update of the agents of the population according to their values and attitudes. For
analytical reasons, we may assume that X is an open interval of R. Under these conditions,
the function Ξ will be differentiable in the first variable.
CMMSE
c 868 ISBN: 978-84-608-6082-2
Consensus formation
The agents interacting through these iterative rules together with the particular parametric
values and initial conditions, will be referred to as a social network. Throughout, we will
suppose that the network is totally connected in the sense that it has no isolated components.
Acknowledgements
This work was partially supported by the National Council for Science and Technology of
Mexico (CONACYT).
References
[1] G. Deffuant, F. Amblard, G. Weisbuch, T. Faure, How can extremism prevail?
A study based on the relative agreement interaction model, Journal of Artificial Societies
and Social Simulation 5 (2002), p.1.
[2] M. H. De Groot, Reaching a Consensus J. Amer. Statist. Assoc. 69 (1974), pp. 118
121.
[3] R. Hegselmann, and U. Krause Opinion dynamics and bounded confidence: models,
analysis and simulation Journal of Artificial Societies and Social Simulation 5 (2002),
p.1.
[5] F.R. Marotto, Introduction to mathematical modeling using discrete dynamical sys-
tems, Thomson Brooks/Cole, London, 2006.
[6] T.J. Scheff, Toward a sociological model of consensus, American Sociological Review
32 (1967), pp. 32–46.
[7] K. Suchecki, V.M. Eguı́luz, and M. San Miguel, Voter model dynamics in com-
plex networks: Role of dimensionality, disorder, and degree distribution, Physical Re-
view E 72 (2005), p. 036132.
[9] T. Yamano, An opinion formation dynamics with logistic map and its complexity,
WSEAS Transactions on Business and Economics 1 (2004), pp.229-234.
CMMSE
c 869 ISBN: 978-84-608-6082-2