Designing Multi-Agent Systems from Logic Specifications - Distributed Artificial Intelligence, Agent Technology, and Collaborative Applications - page 1

Chapter I
Designing Multi-Agent Systems
from Logic Specifications:
A Case Study
Hong Lin
University of Houston—Downtown, USA
aBstract
In this chapter a program construction method based on γ-Calculus is proposed. The problem to be solved
is specified by first-order predicate logic and a semantic verification program is constructed directly from
the specification. We exploit this method in synthesizing the architectural specifications of multi-agent
systems (MAS) in γ-Calculus based on the logic specifications of the MAS. By enabling the transforma-
tion from the logic specifications to operational specifications of MAS, this method allows the design
of the MAS to be focused on the architectural definition level. It benefits the development of MAS by
enabling logic deduction on behaviors of the MAS, and a design methodology in an incremental fashion.
We present this method by a case study of designing a course information management system.
introduction
The modeling issue in the abstract computing machine level has been studied in (Banâtre, Fradet, &
Radenac, 2004), where the chemical reaction model (CRM) (Banatre & Le Metayer, 1990 & 1993;
Banatre, Fradet, & Radenac, 2005a; Le Metayer, 1994) is used to model an autonomic system. Given
the dynamic and concurrent nature of multi-agent systems (MAS), we find that the chemical reaction
metaphor provides a mechanism for describing the overall architecture of the distributed MAS precisely
and concisely, while giving the design of the real system a solid starting point and allowing step-by-step
refinement of the system using transformational methods (Lin, 2004; Lin & Yang, 2006).