How Variables Graphs May Help to Correct Erroneous MAS Specifications

Abstract

This paper is situated in the context of multi-agent systems validation using theorem proving techniques. This document presents a preliminary case study, which is a part of a broader investigation aiming to explore whether such techniques could aid developers in detecting and characterizing errors in MAS specifications. Indeed, regardless of the verification system used (model checking or theorem proving), understanding the reason of a failure in order to correct the specification is most of the time rather difficult. In this article, we propose a method that may help in this task. This method relies on the variables (and their dependencies) that appear in proof obligations generated by GDT4MAS, a specification and verification method dedicated to Multi-Agent Systems. Graphs generated thanks to dependencies between variables occurring in an unproved theorem may indeed help to identify certain types of mistakes, giving a way to correct the specification.