Software system for expert systems learning

Abstract

This paper describes a software system for expert systems learning realized at the School of Electrical Engineering at the University of Belgrade. This software has been developed as visual interactive educational system for students and is aimed to be used within undergraduate and graduate courses. The system guides students through examples and step-by-step tasks related to topics that represent a part of lectures and tutorials. The following topics are covered: breadth-first and depth-first strategies, hill-climbing method, best-first method, branch and bound methods, method A*, production systems, general problem solver, STRIPS planning system and undetermined environment operation systems - fuzzy logic, probability factor reasoning and other ways of expressing uncertainty, ID3 decision tree induction algorithm. Students may enter they own examples and tasks and thus obtain correct solutions. At every stage, it is possible to go backward or forward in the simulation. Student may also print the detailed how-to procedure of solving the task. The implemented system improves lecturer's efficiency and enhances knowledge acquisition of innovative curricula.