On the engineering of agent-based simulations of social activities with social networks

  • Authors:

  • Nicole Ronald;Virginia Dignum;Catholijn Jonker;Theo Arentze;Harry Timmermans

  • Affiliations:

  • Design and Decision Support Systems Group, Eindhoven University of Technology, Eindhoven, The Netherlands;Faculty of Technology, Policy and Management, Delft University of Technology, Delft, The Netherlands;Man-Machine Interaction Group, Delft University of Technology, Delft, The Netherlands;Design and Decision Support Systems Group, Eindhoven University of Technology, Eindhoven, The Netherlands;Design and Decision Support Systems Group, Eindhoven University of Technology, Eindhoven, The Netherlands

  • Venue:
  • Information and Software Technology
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

Context: Models of how people move around cities play a role in making decisions about urban and land-use planning. Previous models have been based on space and time, and have neglected the social aspect of travel. Recent work on agent-based modelling shows promise as a new approach, especially for models with both social and spatial elements. Objective: This paper demonstrates the design and implementation of an agent-based model of social activity generation and scheduling for experimental purposes to explore the effects of social space in addition to physical space. As a side-effect, the paper discusses the need for and requirements on structured design of agent-based models and simulations. Method: Model design was based on the MASQ meta-model and implemented in Python. The model was then tested against several hypotheses with several initial networks. Results: The model allowed us to investigate the effects of social networks. We found that the model was most sensitive to the pair attributes of the network, rather than the global or personal attributes. Conclusion: As demonstrated, a structured approach to model development is important in order to be able to understand and apply the results, and for the model to be extensible in the future. Agent-based modelling approaches allow for inclusion of social elements. For models incorporating social networks, testing the sensitivity to the initial network is important to ensure the model performs as expected.