A simplified protocol operational semantics for multiagent protocols

Multiagent systems are based on the notion that if one gathers simple autonomous systems (agents) within a society and they are endowed with interaction skills, then such a society may show complex behaviors not implemented in any individual (distributed intelligence). Agents interact. The communication issue is typically addressed by means of interaction protocols. Their usual representation by means of graphs only offers a syntactic description of the protocol. No semantic information can be taken into consideration. Moreover, the particular family of protocols that are thus addressed falls in the class of regular languages. This means multiagent systems with such protocols deny themselves any chance of a larger interaction expression power. We go beyond this limit by relying on the protocol operational semantics (POS), inspired by the structural operational semantics of programming languages. It is a formalism based on algebraic data types and pattern-matching which enables to easily describe interaction protocols at an agent level. Such a model is not only a theoretical framework, but also a computational one due to the existence of adequate programming languages.