Title: Acquisition and Exploitation of Gradual Knowledge.
Author: Rose Dieng, Olivier Corby, Stéphane Lapalut
Reference: In International Journal of Human-Computer Studies, Vol. 42, Num. 5, pp. 465-499, 1995.

---

Abstract: Topoi are gradual inference rules, often used by experts in several types of problems: they can be exploited at various phases of a knowledge-based system life cycle. In this article, after justifying the interest of topoi, we try to answer the following questions: how to help the knowledge engineer to elicit gradual knowledge from the expert and to exploit it? In this purpose, we distinguish two viewpoints for studying the notion of topoi, according to the two levels distinguished by Newell: the knowledge level and the symbol level. We formalize topoi through several qualitative physics formalisms. We exploit some knowledge elicitation techniques such as rating grids and some some knowledge acquisition methods such as KADS and KOD in order to facilitate the acquisition of topoi. Last, we study the issues of tooi-based validation and we describe a possible topoi based tool. Throughout the article, we illustrate our ideas by examples of topoi in traffic accident analysis.

---

Keywords: knowledge acquisition, gradual inference, topoi elicitation, qualitative physics.

---