Modeling Coordination in Multi-Agent Systems by Knowledge Diffusion

In this paper we propose a knowledge-based approach for the support of adaptive and context-aware behavior in multi-agent systems. As we will point out, especially the notion of interaction has to be revised in the light of new requirements resulting from a new quality of environmental embedding. In order to provide meaningful and adequate behavior in the presence of dynamic environments systems must possess the ability to interact with the environment on the basis of knowledge about themselves and the situation. We propose the paradigm of knowledge diffusion for the modeling of these coordination processes. We distinguish two extreme forms of knowledge diffusion: indicators support coordination on the basis of uninterpreted signs while ontologies provide semantic support for knowledge-based inferences