2 Model of Interaction Categories

Recent advances in high throughput data acquisition and data storage technologies call for designing distributed agents that are able to learn This paper proposes a new abstract framework for modelling interactions among agents in multi-agent organizations. The proposed model – the model of interaction categories, or MIgories exhibits compositionality of interactions as well as emergence of behavior that is not explicitly designed at the organizational level. The proposed framework is expressive enough to model some of the commonly observed interactions in both natural as well as artificial organizations. More importantly, it allows us to specify and analyze interactions in multi-agent organizations at a fairly high level of abstraction, independent of specific implementations.