Generation of test tasks in the expert-training systems

The article describes methods of generating test tasks that use the knowledge bases of academic disciplines that are formed on the basis of a knowledge production model as well as the apparatus of formal grammars. We present two experimental expert-training systems Formula Tutor and Teoretik, focused mainly on studying the exact sciences. In the Formula Tutor system generation of tasks carried out in two stages: first, by applying the production rules, and then, if necessary, using the outputs in context-free formal grammar. To give tasks to the student familiar forms the apparatus of templates is used. The analysis of student answers is performed by comparing them with the correct answers. If the answer is a mathematical formula it is converted то standard form before comparison. In the Teoretik system production model is also applied. Unlike the first system production rules are used to fix the dependencies of terms of educational discipline. For example, in the course Geometry Ray Dot, Geometric figure shows that to understand the definition of angle the student should know the definition of the ray, dot and geometric figure. The tasks in the system formed by mixing pieces of definitions and provide selectively-constructed answers. Examples in the paper illustrate advantages of described methods.

automated training system, expert-training system, knowledge production model, production rule, context-free formal grammar, task frame, template, selectively-constructed answer