A declarative framework for modeling multi-agent systems

Current work in answer-set programming with regards to its application in the development of reasoning agents has centered around single-agent systems. A well established body of research showing its applicability towards such domains has been developed, describing a thorough methodology for their development upon a theoretical foundation. This work hopes to expand the applicability of this field to the realm of multi-agent domains.

In this work we present a general framework for reasoning about cooperative multi-agent systems. We begin with an overview of the current framework for representing single-agent systems as well as the syntax and semantics of the logic programming language CR-Prolog. Once this baseline has been established, we extend the fundamental notion of an agent to facilitate communication via the introduction of special named sets of fluents known as requests. We then define the notions of an agent's local and global perspectives and their respective diagrams which serve as the theoretical foundation of this work.

Once the general framework has been discussed, a motivating example of a simple multi-agent domain is presented. This example is used to develop a methodology for representing agents capable of reasoning in such domains using the logic programming language of CR-Prolog, together with an axiomatization of multi-agent communication.

Finally a series of results detailing some fundamental properties of the framework are presented.