SeXAI: Introducing Concepts into Black Boxes for Explainable Artificial Intelligence

The interest in Explainable Artificial Intelligence (XAI) research is dramatically grown during the last few years. The main reason is the need of having systems that beyond being effective are also able to describe how a certain output has been obtained and to present such a description in a comprehensive manner with respect to the target users. A promising research direction making black boxes more transparent is the exploitation of semantic information. Such information can be exploited from different perspectives in order to provide a more comprehensive and interpretable representation of AI models. In this paper, we present the first version of SeXAI, a semantic-based explainable framework aiming to exploit semantic information for making black boxes more transparent. After a theoretical discussion, we show how this research direction is suitable and worthy of investigation by showing its application to a real-world use case.