Social Space in Simulation Models

Social Space in Simulation Models

Davide Nunes (University of Lisbon, Portugal) and Luis Antunes (University of Lisbon, Portugal)
DOI: 10.4018/978-1-4666-5954-4.ch012
OnDemand PDF Download:
List Price: $37.50


In real world scenarios, the formation of consensus is a self-organisation process by which actors have to make a joint assessment about a target subject, be it a decision making problem or the formation of a collective opinion. In social simulation, models of opinion dynamics tackle the opinion formation phenomena. These models try to make an assessment, for instance, of the ideal conditions that lead an interacting group of agents to opinion consensus, polarisation or fragmentation. This chapter investigates the role of social relation structure in opinion dynamics and consensus formation. The authors present an agent-based model that defines social relations as multiple concomitant social networks and explore multiple interaction games in this structural set-up. They discuss the influence of complex social network topologies where actors interact in multiple distinct networks. The chapter builds on previous work about social space design with multiple social relations to determine the influence of such complex social structures in a process such as opinion formation.
Chapter Preview


In social simulation, the way one defines social spaces is determinant to construct relevant scenarios to target problems. Moreover, such spaces have a deep influence in results of the simulations by imposing restrictions that guide or filter agent interactions. In this chapter, we discuss the connection between models of social structures and phenomena observed in simulations at a macro level. We conduct a series of experiments to highlight the relevance of proper social space design considerations, when creating social simulation models.

In agent-based social simulation models, the way we deal with social structure varies according to multiple factors. These models may span across multiple levels of abstraction and are usually targeted for specific phenomena in order to shed light on some complex process that follows a set of observed properties. Examples of social space models include: the usage of discrete grids or lattices in which agent interaction is restricted to certain neighbourhoods. Cellular automata (CA) models make use of such regular grids and are increasingly used to study a variety of social dynamics (Flache & Hegselmann, 2001). Other common approach, is the usage of generative procedures to create random network structures exhibiting real-world properties such as the celebrated Barabási-Albert model of preferential attachment (Barabási & Albert, 1999). The preferential attachment mechanism generates network models with scale-free properties, these properties can be observed in real-world structures such as the World Wide Web (Barabási, Albert, & Jeong, 2000). Other existing models try to describe aspects of real-world social networks. The work in (Hamill & Gilbert, 2009) for instance, gets inspiration from the notion of social circles to model key aspects of large real social networks such as low density, high clustering and preferential-attachment-like behaviour of degree of connectivity.

In this chapter we aim to present and discuss available modelling methodologies to design social spaces for simulation models; we also discuss the need for more complex social space designs, presenting our approach to model multiple concurrent social relations; finally we aim to discuss the influence of social networks in self-organisation phenomena such as consensus and opinion formation. We describe a model of opinion dynamics with multiple social networks and compare our work with existing literature on opinion dynamics and diffusion models.

One of the main subjects of this chapter is the study of models of social networks and how they affect social phenomena. It is important to notice that the study of models of social structure and their impact on simulation models is valuable not only as a contribution to complexity sciences, but also to develop new managerial tools and systems that can be used to deal with real problems. Complexity theory didn't deliver a general theory of complex organisations yet, but the efforts in the last couple of decades generated many useful components such as modelling methodologies and simulation at different levels of abstraction. While these tools do not capture the complete set of properties from the world they mirror, the study of hierarchical structures, networks of interaction, non-linearity and emergence (see chapter 1) generated useful and in some cases practical understanding about observable properties (Cilliers, 2001). Simulation in general and social simulation in particular have allowed us to build bridges between the micro components of complex systems (that we know as facts or conjecture about its existence) and the macro effects of the processes being studied.

Complete Chapter List

Search this Book: