Knowledge Source Confidence Measure Applied to a Rule-Based Recognition System

Paper deals with the knowledge acquisition process for the design of the decision support system. Usually in this case the knowledge is given in the form of rules which are formulated by human experts or/and generated on the basis of datasets. Each of experts has different knowledge about the problem under consideration and rules formulated by them have different qualities. The qualities of data stored in the databases are different as well. It might cause differences in quality of generated rules. In the paper we formulate the proposition of a knowledge source confidence measure and we show some of its applications to the decision process e. g., we show how to use it for contradiction elimination in the set of rule. Additionally, we propose how it could be used during decision making.