Using Conceptual Graphs to represent Problem-Solving Methods

摘要

Modeling a Problem-Solving Method (PSM) within the Task/Method paradigm consists in representing problems to be solved as tasks, and the ways to solve a problem as methods. The PSM is then applied to real cases by iterating the following steps: select, according to the current solving context, the more relevant task to be achieved next and, then, select the more pertinent method to be performed to achieve the previously selected task. The construction of a Task/Method knowledge base also requires the mo-delisation of knowledge related to the domain on which the considered PSM will be applied. For this purpose, different knowledge representation languages have been proposed such as logic or semantic networks. In this paper, we present a new model dedicated to the explicitation of a PSM in the context of the Task/Method paradigm. This model aims at taking advantage of research related to knowledge representation and PSM modeling. Our approach consists in using Conceptual Graphs to represent the considered domain, the tasks and the methods, and to define the selection/exploitation mechanisms. A predefined Task/Method engine which allows a dynamic selection of tasks and methods is also presented.