Behavior-Driven Explainability: A Specification-Based Approach to Trustworthy System Design
As system complexity continues to grow, it has become increasingly difficult for individuals or teams to fully comprehend every aspect of a modern system throughout its development lifecycle—from design through maintenance. This challenge is especially critical for safety-critical systems, where trust in the final design is paramount. In response, explainability has emerged as a key requirement. This paper introduces Behavior-Driven Explainability (BDX), a novel method that leverages the principles of Behavior-Driven Development (BDD) to derive clear, structured explanations from system specifications.
BDD describes expected system behavior through structured scenarios—sequences of actions for each functionality—which can be directly translated into human-readable explanations. BDX applies this concept to provide explanations at any stage of development or abstraction level. To demonstrate its practical value, the authors present a case study on a RISC-V processor, where BDX explains exceptions during system design, illustrating how this approach enhances trust and understanding in complex, safety-critical systems.
Key Contributions
- Introduces the BDX methodology, bridging BDD scenarios with explainable AI principles.
- Demonstrates applicability across development stages, from early design to maintenance.
- Provides a concrete case study for exception handling in RISC-V processors, a domain where transparent reasoning is critical.
Subjects
- Primary: Machine Learning (cs.LG)
- Related: Hardware Architecture (cs.AR); Software Engineering (cs.SE)
ACM Classification
- D.2.1 (Software Engineering – Requirements/Specifications)
- D.2.3 (Software Engineering – Coding Tools and Techniques)
- D.2.4 (Software Engineering – Software/Program Verification)
- B.5.2 (Hardware – Register-Transfer-Level Implementation)
- B.8.2 (Performance and Reliability – Performance Analysis and Design Aids)
Published: arXiv:2607.24881 (cs.LG), Submitted July 27, 2026
via ArXiv LG