Syntax-Directed Translation Schemes for Multi-Agent Systems Conversation Modelling

In modern organisations the monolithic information systems of the past are being gradually replaced by networked systems, enabling distributed computing often based on multi-agent system architectures. This new paradigm enables the use of information systems support in new areas of organisational activity, especially those involving the interaction of business agents. All communication-intensive business processes based on formal conversations, i.e. partially ordered sets of communicative acts transmitted among a set of agents, qualify as candidates to, at least partial, automation. Still a very active area of research, this paradigm has been studied in areas such as distributed artificial intelligence, organisational simulation and workflow management. However, in all these areas the basic problem is the adequate representation of agent conversations. In this paper we present a formal method for conversation representation that is inspired in syntactic pattern recognition methods, specifically syntax-directed translation schemes. This method has a clear semantics that can be easily given a declarative implementation, thus becoming flexible enough to accommodate on-line extensions and exception handling.