Vers une formulation floue des explications par contraste


Explaining a decision requires some properties that have been studied and established in cognitive sciences. An important one is the contrastive component : an explanation should allow answering questions such as “why make decision P rather than Q?”. This principle has been formalized recently by Miller in a logical framework exploiting knowledge represented as causal graphs with variables taking crisp values. This paper proposes to explore several directions for integrating fuzzy knowledge or data : a situation can e.g. be described with fuzzy terms ; likewise a cause or an explanation can be expressed using linguistic modalities, contributing to the explanation simplicity.