Knowledge representation and human-computer interaction in an intelligent tutor for diagnostic problem solving

Implementation of an intelligent tutor for diagnostic problem solving in a complex engineering domain requires: (1) a large amount of domain knowledge organized to facilitate evolution of system states with time, (2) troubleshooting task knowledge, (3) knowledge to infer a student's possible misconceptions from observed actions, and (4) pedagogical knowledge to realize the tutoring objectives. The domain and task knowledge needs to be organized in a manner that is easily accessible and communicable to the student, and the instructional system requires a well designed interface to communicate this knowledge. A methodology for representing domain and pedagogical knowledge in an intelligent tutoring system is described. An implementation of the methodology in an instructional system that can train operators to troubleshoot a marine power plant is described.<
>