Contrastive and counterfactual explanations for test case prioritization: Ideas and challenges

• Aurora Ramírez ^[1] ; Mario Berrios ^[1] ; Robert Feldt ^[2] ; José Raúl Romero ^[1]
  1. [1] Universidad de Córdoba

     Universidad de Córdoba

     Cordoba, España

     
  2. [2] Chalmers University of Technology

     Chalmers University of Technology

     Suecia

     
• Localización: Actas de las XXVII Jornadas de Ingeniería del Software y Bases de Datos (JISBD 2023) / coord. por Amador Durán Toro , 2023
• Idioma: inglés
• Enlaces
  • Texto completo
• Resumen

  • As machine learning (ML) is increasingly used in software engineering (SE), explainable artificial intelligence (XAI) is crucial for understanding choices made by opaque, "black-box" models. Test case prioritization (TCP) is an important SE problem that can benefit from ML. In this paper, we explore two approaches for generating explanations in ML-based TCP, contrastive and counterfactual XAI, and present application scenarios where they can enhance testers' comprehension of model outputs. Specifically, we use DiCE, a method for generating counterfactual explanations, as an illustrative example and conclude by discussing open issues.