• 2010
    • Investigating new social algorithms and hybrid evolutionary agent models for ABSS (Agent-Based Social Simulations).
      (Marcelo Pita)

      The simulation results of the research which vary from the instantiation of simplified environments to real scenarios that reflect little towns have used real data from public health care databases and reveal interesting population dynamics for specific contexts strengthening the initial hypothesis of applicability of the models and tools to governmental decision support systems.

      It was proposed a new hybrid evolutionary agent model that incorporates two intelligent decision components: (a) subsymbolic - responsible by the codification and evolutionary dynamics of behaviors and characteristics genetically motivated in agents (theoretically based on the evolutionary theory of Charles Darwin) - and (b) symbolic - responsible by the learning and cultural evolutionary dynamics in agents (theoretically inspired by the theory of memes of Richard Dawkins). Additionally it was proposed a new model and an implementation (PAX framework) for structuring social environment where not only behavioral and temporal relationships among entities of the environment are considered but also the spatial ones.

    • Support Decision Tool Using Agent Based Social Simulation as Engine
      (Marcos Álvares)

      This research aims the creation of a support decision tools' architecture using agent based social simulation as engine. This tool will support public officers on high impact decisions generating future scenarios from macroeconomic input data. A multi-objective genetic algorithm will be used to select these scenarios according social and economic aspects.

      The output will be a set of future scenarios (with actions to achieve each one). The idea is offer "suggestions" to support the public officer decision conciliating concurrent objectives.

    • Framework for Social Simulation Aspect-Oriented
      (Diego Siqueira)

      Social Simulation is an area of research that uses computational methods to solve problems in social sciences (politics economics anthropology etc.).

      The tools available today to perform social simulations of agent-based models require considerable effort for modeling as they still require non-trivial skills from the user to define new types of simulations.

      In order to simplify the implementation of social simulations using agents this work proposes a framework for social simulations aspect-oriented to describe more friendly simulations making them semi-transparent to the user the details of computational and artificial intelligence techniques used.