WER2026 - 29th Workshop on Requirements Engineering


Explainability Requirements Engineering in AI-enabled Systems

Lívia Mancine; Renata Braga; Renato Bulcão-Neto

10.29327/1884517.29-25

PDF Scholar

Abstract

Explainability is the ability to make the behavior and decisions of AI-enabled Systems understandable to stakeholders and has become essential for trust, transparency, and informed decision-making across application domains. However, Requirements Engineering (RE) still lacks systematic approaches for integrating explainability throughout the development lifecycle in a structured and stakeholder-oriented manner. This doctoral research proposes OpenUPExp, a process-oriented RE framework that extends the Open Unified Process by introducing roles, activities, and artifacts to support explainability by design as a non-functional requirement. The research follows the Design Science Research methodology and evaluates the framework through illustrative scenarios and real-world case studies involving AI-enabled Systems. The main contribution is OpenUPExp, a process-oriented RE approach for the systematic treatment of explainability from conception through elicitation, analysis, and specification. The framework aims to support stakeholder-centered explainability, improve the traceability and documentation of explainability requirements, and facilitate their alignment with different stakeholder needs and contexts.

Keywords: Requirements Engineering; Explainability; AI-enabled Systems; OpenUP; Design Science Research