Cooperating logical agents

Abstract

The paper presents a logic-based model of Multi-Agent Systems with the double aim to be a framework for design and implementation and a tool in theoretical study of such systems. In particular we are interested in the problems raised by conceptual heterogeneity of agents in concrete Multi-Agent Systems, and in giving a “grounded” semantics of mental attitudes, based on their cognitive activity. First we present the model itself, in which agents are higher-order logic processes, with complex reasoning and computing abilities, which communicate by message-passing. Secondly this model is used as a formal basis for analysing Belief and Knowledge. Our approach is of the “deductive” kind: to believe P is basically to have (or potentially have if implicit belief is concerned) a proof for P. This can be conveniently formalised in the model using notions and techniques from proof theory, rather than from model theory as in both the standard “modal” and “syntactic” approaches, and from higher-order rather than from modal or first-order logic. Then we discuss the problems arising from communication and cooperation between heterogeneous agents. Finally, we shortly present a partial implementation of the model and its application in office automation.