Expert systems are popular ways to solve very complex and hard problems. However, it is well-known that knowledge acquisition is a bottleneck process to develop them. Furthermore, the development of the systems can fail because there is no expert or an expert less qualified.
In order to overcome the problems that they possess, this thesis focuses on an extended architecture of the expert systems. A simulator and an induction system are added to the existing architecture of expert systems. An expert system for schedule-based material requirements planning(SBMRP) has been implemented to show how the extended architecture works, and produces better results than existing SBMRP systems.

Abstract
 1. 서론
 2. 지식획득 과정
  2.1 진정한 전문가가 존재할 경우
  2.2 전문가가 존재하나 성능개선의 여지가 존재하는 경우
  2.3 전문가가 존재하지 않은 경우
 3. 제안된 지식획득 구조
  3.1 일반화된 구조
  3.2 예제
 4. 결론
 참고문헌

• 김선욱(단국대학교 산업공학과)