Modeling and verifying context-aware non-monotonic reasoning agents

Abstract

This paper complements our previous work on formal modeling of resource-bounded context-aware systems, which handle inconsistent context information using defeasible reasoning, by focusing on automated analysis and verification. A case study demonstrates how model checking techniques can be used to formally analyze quantitative and qualitative properties of a context-aware system based on message passing among agents. The behavior (semantics) of the system is modeled by a term rewriting system and the desired properties are expressed as LTL formulas. The Maude LTL model checker is used to perform automated analysis of the system and verify non-conflicting context information guarantees it provides.