Norm Monitoring Under Partial Action Observability

Area of Science:

• Artificial Intelligence
• Computer Science
• Multiagent Systems

Background:

• Norms are crucial for controlling multiagent systems.
• Monitoring norm compliance is challenging under partial action observability.
• Existing literature lacks mechanisms for this specific monitoring problem.

Purpose of the Study:

• To address the problem of monitoring norm compliance in multiagent systems with partial action observability.
• To propose a novel approach for reconstructing unobserved actions to enable norm monitoring.
• To develop and evaluate algorithms for norm monitoring under these conditions.

Main Methods:

• Formalizing the problem of reconstructing unobserved actions.
• Developing an information model for partial action observability.
• Proposing algorithms utilizing two distinct reconstruction processes.
• Conducting evaluations to assess the effectiveness of the proposed methods.

Main Results:

• The proposed reconstruction of unobserved actions significantly increases the detection of norm violations.
• The method also significantly enhances the detection of norm fulfillments.
• The developed algorithms provide a viable solution for norm monitoring under partial observability.

Conclusions:

• Reconstructing unobserved actions is an effective strategy for monitoring norm compliance in partially observable multiagent systems.
• The proposed information model and algorithms offer a practical solution to a previously unaddressed challenge.
• This work advances the field of norm-based control in multiagent systems.