Requirements-Based Explainability for Multi-agent Systems

doi:10.1007/978-981-95-4969-6_19

Requirements-Based Explainability for Multi-agent Systems

2026-01-01

1

PRE

AI

S

Sebastián Rodríguez *

DOI：10.1007/978-981-95-4969-6_19

Original

Original request for help

Save

Abstract

En 中文

Explainability is key to building trust in AI systems, especially in autonomous multi-agent decision-making. While many techniques exist, full trust is only achieved when explanations clarify decision criteria and trace them back to original requirements. We extend explainable agents (XAg) by linking decisions to user and system stories, along with their acceptance criteria. This approach ensures full traceability, revealing decision rationale and aligning them with origins. It also allows runtime verification of decisions, flagging errors, and supporting requirement reflection if poor decisions are made.

Requirements-Based Explainability for Multi-agent Systems

Abstract

Journal

Organization