ABSTRACT
Explainability has a place among the assistance attributes: An assistant system can explain its functionality and behavior. Explanations are in general reactions on specific questions that refer to a lack of knowledge. An explanation of how a problem has been solved can be given for a complete solution, for a partial solution, or for a part of a complete or partial solution. The first kind of explanation seems to be especially difficult because the enormous amount of information in a complete solution trace is far too detailed to be an informative explanation for the user. The explanation ability is one of the basic properties a knowledge-based system implementing assistant behavior must have. After positioning explainability into the context of the assisting computer paradigm, experiences from the project DIAMOD have been reported, going into some detail with regard to KADS, the knowledge representation and acquisition technique used there, and its advantages for explanation purposes.
