Using Case-Based Reasoning for Capturing Expert Knowledge on Explanation Methods

Model-agnostic methods in eXplainable Artificial Intelligence (XAI) propose isolating the explanation system from the AI model architecture, typically Machine Learning or black-box models. Existing XAI libraries offer a good number of explanation methods, that are reusable for different domains and models, with different choices of parameters. However, it is not clear what would be a good explainer for a given situation, domain, AI model, and user preferences. The choice of a proper explanation method is a complex decision-making process itself. In this paper, we propose applying Case-Based Reasoning (CBR) to support this task by capturing the user preferences about explanation results into a case base. Then, we define the corresponding CBR process to help retrieve a suitable explainer from a catalogue made of existing XAI libraries. Our hypothesis is that CBR helps the task of learning from the explanation experiences and will help to retrieve explainers for other similar scenarios.