Modelling from knowledge versus modelling from rules using UML

Abstract

Modelling support for knowledge acquisition is a tool for modelling domain knowledge. However, during the implementation of the knowledge new knowledge is created. Event though this knowledge is found in the knowledge base, the model usually is not updated with the new knowledge and do, therefore, not contain all the knowledge in the system. This paper describes how different graphical models support the complex knowledge acquisition process of handling domain knowledge and how these models can be extended by modelling knowledge from rules in a knowledge base including probability. Thus, the models are designed from domain knowledge to create production rules but the models are also extended with new generated knowledge, i.e., generated rules. The paper also describes how different models can support the domain expert to grasp this new generated knowledge and to understand the uncertainty calculated from rules during consultation. To this objective, graphic representation and visualisation is used as modelling support through the use of diagrams of Unified Modelling Language (UML), which is used for modelling production rules. Presenting rules in a static model can make the contents more comprehensible and in a dynamic model can make the uncertainty more evident. © Springer-Verlag Berlin Heidelberg 2005.