MAS as Complex Systems: A View on the Role of Declarative Approaches

摘要

The ever growing complexity of software systems calls for new forms of understanding and conceptual tools. It has been argued that some 'Laws of Complexity' exist, which govern the behaviour of complex systems of any sort, from natural to artificial ones. Along this line, in this paper we draw from the most recent findings of evolutionary biology to develop an original view over Multiagent Systems (MAS). A schema for a "layered", hierarchical view of MAS is introduced, aimed at providing computer scientists and engineers with a powerful conceptual framework for MAS observation / modelling / construction. We first introduce the three levels of the hierarchy in general, and then show how they impact on current proposals for methodologies of agent-oriented analysis and design. Finally, we exploit the hierarchy to provide an overall organised view over declarative approaches to MAS, by using as a reference those presented in the other contributions in this book. On the one hand, a hierarchical view allows the many different approaches to be distinguished, classified and possibly compared. On the other hand, it makes it possible to reveal the richness and diversity of declarative models and technologies for MAS, as well as to show the multiplicity of ways in which they impact on MAS modelling and engineering.