On the external concurrency of current BDI frameworks for MAS

2024-04-16

International Workshop on Engineering Multi-Agent Systems

AI Summary

BDI agents in multi-agent systems (MAS) exhibit inconsistent and race-prone responses to external concurrent events. Method: We propose the first evaluation paradigm for BDI frameworks under external concurrency, combining formal modeling, concurrent event injection, temporal logic verification, and empirical simulation across multiple frameworks (e.g., Jason, 2APL). Contribution/Results: Our analysis systematically identifies 12 classes of concurrency anomalies across five mainstream BDI frameworks, exposing critical coordination failures among event scheduling, belief update, and intention execution. Based on these findings, we design a lightweight intervention protocol that preserves original framework architectures while significantly improving event-processing correctness—from an average of 63% to 91%. The protocol further enhances system stability and result reproducibility, offering a practical, deployable solution to concurrency-related reliability issues in BDI-based MAS.

Technology Category

Multiagent Systems: Agent-Based Simulation and Emergent BehaviorCognitive Modeling & Cognitive Systems: Agent ArchitecturesKnowledge Representation and Reasoning: Reasoning with Beliefs

Application Category

Search and Retrieval-Augmented AI: Web evaluation methodologies and metricsUser Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systemsSemantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Problem

• Multi-Agent Systems
• Parallel Processing
• BDI Agents Optimization

Innovation

• Multi-Agent Systems
• Parallel Processing Strategies
• Behavioral Design Flexibility