Abstract:

Software systems for knowledge testing contain various standard types of questions and tasks. Formal construction and internal implementation of individual types does not give the user/teacher much freedom. When creating a database of questions and tasks, they must follow the fixed rules. Thus, he must formulate individual tasks and solutions as required by the system. The same applies to the evaluation of correct answers and solutions. For many scientific disciplines, especially in the field of exact sciences, where we expect a creative approach to the solution, this can mean an unpleasant limitation. Another limiting factor is when the system recognizes only the correct solution and only a small deviation from the expected solution is evaluated as an incorrect answer. It is also unpleasant when the system does not allow to evaluate the partial correctness of the solution, e.g. used correct procedure with small numerical error.

Systems for examining and knowledge testing only exceptionally include the possibility of using parametric tasks. At the same time, tasks of this type are very useful and often used in exact scientific disciplines such as mathematics, physics, chemistry, computer science, electronics, engineering, and economics, etc. The natural equivalent of parametric problems in classical education are tasks defined in text form. This type of task can be solved by different values of parameters. In this case, the learners solve equally challenging tasks. The author of the test tasks easily covers the chosen thematic unit both for practicing and fixing the knowledge, as well as for checking the depth and quality of the acquired knowledge. The paper describes the possibilities of using standard types of test tasks with a focus mainly on parametric tasks in various subjects. It presents a number of examples (mainly from mathematics and computer science), analyzes the didactic value of some of the examples and their significance in the educational process. It also points out the connections between parametric tasks, learning with understanding and deep learning.

Keywords:

Testing, classification, knowledge testing, parametric tasks, computer-aided algorithm-driven testing.