Method of Test Generation for Debugging of Knowledge Bases for Artificial Intelligence Systems

Dolinina O. N., e-mail: odolinina09@gmail.com

There are considered main types of errors in knowledge bases of expert systems, based on rules and on artificial neural networks. There are described methods of static and dynamic debugging of knowledge bases. Method of building of full set of tests for knowledge bases is suggested. There is described its using for testing of rule-based systems and for artificial intelligence systems based on 3-level perceptron.

Keywords: artificial neural networks, rule-based systems, errors in knowledge bases, debugging of expert systems, type of error "forgetting about exception", knowledge extraction, logic schemes

pp. 40–47