A fully abstract model for the exchange of information in multi-agent systems

In this paper, we present a semantic theory for the exchange of information in multi-agent systems. We consider the multi-agent programming language agent communication programming language, which integrates the paradigms of concurrent constraint programming and communicating sequential processes (CSP). The constraint programming techniques are used to represent and process information, whereas the synchronous communication mechanism from CSP is generalised to enable the exchange of information. The semantics of the language, which is based on a generalisation of traditional failure semantics, is shown to be fully abstract with respect to observing of each terminating computation its <nal global store of information. c © 2002 Elsevier Science B.V. All rights reserved.