The Trainer System: Applying QR Techniques to Intelligent Tutoring Systems

Publisher Summary This chapter proposes a design framework for an intelligent training system (ITS), called TRAINER, in the engineering domain. TRAINER uses both implicit and explicit domain knowledge representations in the teaching of control concepts in dynamic physical systems. Implicit representation in TRAINER is formulated by pairs of if-then rules to decide on particular model(s) to fire depending on the underlying assumptions and the working conditions of the plant. Explicit representation based on qualitative reasoning (QR) techniques then focuses on the activated model through the process known as tracking in order to infer the trainee's knowledge state. In engineering domains, where knowledge of the internal structure already exists, the explicit knowledge approach is particularly fruitful. This internal representation extends the range of operability of the implicit ES system representation. This chapter elucidates a preliminary design of the TRAINER system as well as some results from a formative evaluation. One main line of research toward the representation of explicit models is the study of causal qualitative models and the respective causal QR techniques. This chapter presents research that explores the use of QR techniques in the design and development of ITSs in the engineering domains.