Knowledge acquisition for explainable, multi-expert, knowledge-based design systems [chapter]

Rose Dieng, Alain Giboin, Paul-André Tourtier, Olivier Corby
1992 Lecture Notes in Computer Science  
In order to help the knowledge engineer and the expert during knowledge acquisition phase, the ACACIA Group is working on a knowledge acquisition methodology and tool (KATEMES) allowing knowledge acquisition from multiple experts, exploiting the specificities of design problems and preparing the assistance to the end-user and the quality of explanations he will be provided with. This paper describes our research program. After a brief description of our previous knowledge acquisition tool
more » ... we will present the primitives of KATEMES and the problems we intend to study and the ideas we intend to deepen about the link between knowledge acquisition and explanations, knowledge acquisition from multiple experts and methodological aspects. Keywords: knowledge acquisition and explanations, knowledge acquisition from multiple experts, knowledge graphs, cognitive agents, design applications 299 had a number of experiments in the past [59, 58, 60] . Notice that very few knowledge acquisition tools concern design task, that is considered as very difficult to tackle: DSPL ACQUIRER [13] , SALT[51], CANARD [69]. Here are the main questions we will study: 1) How to elicit and model knowledge from several experts? 2) How to take into account the users of the final knowledgebased system (KBS) and, in particular, to prepare the quality of the future explanations in the earlier knowledge acquisition phase? 3) How to validate the acquired knowledge? 4) How to design the final multi-experts system? 5) Can we propose a methodology taking into account explanations and multi-expertise? KATEMES (Knowledge Acquisition Toot for Explainable, Multi-Expert Systems) will be the result of this research. This paper will describe our research program. After a brief description of our previous knowledge acquisition tool 3DKAT, we will present the primitives of KATEMES and the problems we intend to study and the ideas we intend to deepen about the link between knowledge acquisition and explanations, knowledge acquisition from multiple experts and methodological aspects. Very-long-term research topics such as validation of the acquired knowledge and design of the final system will also be mentioned in order to give a complete idea of the ACACIA program. At the end of each section, we indicate related work, in order to recapitulate the points for which similar ideas were already evoked by other researchers and to highlight those for which we propose an original approach.
doi:10.1007/3-540-55546-3_47 fatcat:2iawzsmyfjeujcd7dejhtjagzq