[Paper Review] Verification of Agent-Based Artifact Systems

Francesco Belardinelli, Alessio Lomuscio+1 more|arXiv (Cornell University)|Jan 12, 2013

Business Process Modeling and Analysis32 references4 citations

TL;DR

This paper proposes a formal framework for verifying agent-based artifact-centric systems using a novel semantics called artifact-centric multi-agent systems (AC-MAS), which integrates data structures and agent interactions. It identifies two decidable fragments—via syntactic restriction to sentence-atomic formulas or semantic uniformity—enabling finite abstractions and proving the model checking problem to be EXPSPACE-complete, thus enabling verification of infinite-state systems through finite models.

ABSTRACT

Artifact systems are a novel paradigm for specifying and implementing business processes described in terms of interacting modules called artifacts. Artifacts consist of data and lifecycles, accounting respectively for the relational structure of the artifacts' states and their possible evolutions over time. In this paper we put forward artifact-centric multi-agent systems, a novel formalisation of artifact systems in the context of multi-agent systems operating on them. Differently from the usual process-based models of services, the semantics we give explicitly accounts for the data structures on which artifact systems are defined. We study the model checking problem for artifact-centric multi-agent systems against specifications written in a quantified version of temporal-epistemic logic expressing the knowledge of the agents in the exchange. We begin by noting that the problem is undecidable in general. We then identify two noteworthy restrictions, one syntactical and one semantical, that enable us to find bisimilar finite abstractions and therefore reduce the model checking problem to the instance on finite models. Under these assumptions we show that the model checking problem for these systems is EXPSPACE-complete. We then introduce artifact-centric programs, compact and declarative representations of the programs governing both the artifact system and the agents. We show that, while these in principle generate infinite-state systems, under natural conditions their verification problem can be solved on finite abstractions that can be effectively computed from the programs. Finally we exemplify the theoretical results of the paper through a mainstream procurement scenario from the artifact systems literature.

Motivation & Objective

• To address the lack of formal verification methods for multi-agent systems operating on artifact-centric workflows that combine data and process.
• To formalize artifact-centric multi-agent systems (AC-MAS) with a computationally grounded semantics that captures both data evolution and agent knowledge.
• To identify decidable fragments of the model checking problem for first-order temporal-epistemic logic in infinite-state artifact systems.
• To develop a constructive method for building finite abstractions from artifact-centric programs, enabling effective verification.
• To demonstrate the approach on a real-world procurement scenario, showing practical applicability and correctness.

Proposed method

• Proposes AC-MAS as a formal model that combines artifact systems with multi-agent systems, explicitly modeling data structures and agent interactions.
• Introduces a first-order temporal-epistemic logic with quantifiers to express agent knowledge and system evolution over time.
• Identifies two orthogonal decidable fragments: one via syntactic restriction to sentence-atomic formulas, the other via semantic uniformity of the system.
• Develops novel notions of first-order bisimulation to construct finite abstractions that preserve logical properties.
• Defines artifact-centric programs as declarative, compact representations of system behavior, showing they induce uniform AC-MAS under bounded conditions.
• Provides a constructive procedure to generate finite abstractions from these programs, enabling model checking on finite models.

Experimental results

Research questions

• RQ1 Can model checking be made decidable for infinite-state artifact-centric multi-agent systems with first-order data and epistemic logic?
• RQ2 What syntactic or semantic restrictions allow finite abstractions of infinite-state artifact systems while preserving logical properties?
• RQ3 Is the uniformity condition sufficient for finite abstraction, and can it be applied to general artifact-centric programming languages?
• RQ4 Can artifact-centric programs be used to generate finite abstractions that preserve verification results for complex specifications?
• RQ5 How does the computational complexity of model checking scale under these restrictions, and is it tractable in practice?

Key findings

• The model checking problem for artifact-centric multi-agent systems against first-order temporal-epistemic logic is undecidable in general.
• Under the syntactic restriction to sentence-atomic formulas, infinite-state bounded AC-MAS admit finite abstractions via first-order bisimulation.
• Under the semantic condition of uniformity, bounded AC-MAS also admit finite abstractions, enabling decidable verification.
• The model checking problem for these decidable fragments is EXPSPACE-complete, indicating high but decidable complexity.
• The methodology was successfully applied to a procurement scenario, verifying properties like matching, budget constraints, and cost limits, while correctly rejecting fulfillment requirements.
• Finite abstractions were constructed from artifact-centric programs, and model checking on these finite models correctly reflected the truth of specifications on the original infinite system.