A User-Centered Ontology for Explainable Artificial Intelligence in Aviation

Denys Bernard, Jaume Perello-March, Ignacio Solis-Marcos, Martine Cooper

2024

Abstract

Aviation authorities will require future decision assistance systems based on Artificial Intelligence (AI) to be explainable in order to enhance trust, safety, situation awareness (SA) and to promote appropriate use of the system. We anticipate that authoring implementable explainability requirements will be a challenge for the relevant stakeholders in the aerospace industry. Here we propose an ontology for explainable AI (XAI) from a user centered perspective in aviation. We propose that the development of an adequate mental model of XAI has to be considered as a naturally dialogic process between the user and the AI, where the need for an explanation can be approached as a question. The explanation specification process describes the informational content of explanations,, whose main components are an Explanans and an Explanandum linked by the appropriate discourse relation. The Explanandum denotes the aspect of the outcome of the system about which the operator needs an explanation, and the Explanans is typically a set of true facts which actually satisfy the operator's cognitive need. We understand explanation as a communication act with the purpose of making the user accept or better understand the Explanandum. Thus, an explanation is successful only if a particular relationship holds between the explanans and the explanandum. To understand such a relationship, we refer to the theory of discourse where the so-called discourse relations act as the logical core of a discourse, and are constitutive of its consistency. By using this ontology, aviation Human Factors and operations practitioners will be able to specify the content of explanations in order to maximize the acceptability and usability of explanations by the user.

Paper Citation

in Harvard Style

Bernard D., Perello-March J., Solis-Marcos I. and Cooper M. (2024). A User-Centered Ontology for Explainable Artificial Intelligence in Aviation. In Proceedings of the 2nd International Conference on Cognitive Aircraft Systems - Volume 1: ICCAS; ISBN 978-989-758-724-5, SciTePress, pages 88-94. DOI: 10.5220/0012978600004562

in Bibtex Style

@conference{iccas24,
author={Denys Bernard and Jaume Perello-March and Ignacio Solis-Marcos and Martine Cooper},
title={A User-Centered Ontology for Explainable Artificial Intelligence in Aviation},
booktitle={Proceedings of the 2nd International Conference on Cognitive Aircraft Systems - Volume 1: ICCAS},
year={2024},
pages={88-94},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0012978600004562},
isbn={978-989-758-724-5},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 2nd International Conference on Cognitive Aircraft Systems - Volume 1: ICCAS
TI - A User-Centered Ontology for Explainable Artificial Intelligence in Aviation
SN - 978-989-758-724-5
AU - Bernard D.
AU - Perello-March J.
AU - Solis-Marcos I.
AU - Cooper M.
PY - 2024
SP - 88
EP - 94
DO - 10.5220/0012978600004562
PB - SciTePress