Applying Logic Programming to theSpeci cation of Complex

In this paper we show how multi{theory meta{logic programming techniques can be applied to the realization of multi{agent systems which solve real{world complex problems, in which the integration of heterogeneous software environments could be necessary. We have deened a language named ACLPL (i.e., Agent Constraint Logic Programming Language) implemented in the constraint logic programming language Eclipse and extending standard (constraint) logic programming. ACLPL provides an environment in which the global knowledge is partitioned into theories (i.e., agents) and also primitives for communication among agents, updating of an agent's knowledge base and simulation of the execution of a multi{agent system. Our nal aim is to realize a speciication tool for multi{agents systems using logic programming techniques as well as software engineering ones. At the moment, the approach we use to obtain an executable speciication is simple: we identify the set of agents the application needs and give a high{level informal description of the interactions among agents, then we implement each agent by means of a diierent logical theory, translating the static speciication (given by a transition function describing the behaviour of the agent) into ACLPL. Finally we execute the obtained system, to test the implementation choices. As a demonstration of our approach we present a planner for goods transportation: four kinds of agents, Client, Agency, Distributor and Transporter interact to plan the delivery of goods from a place to another.