Associative Methods as a Tool to Improve the Quality of Knowledge Control

Abstract

One of the tools to control the level of knowledge is testing. When compiling a control test, it is necessary to ensure full coverage of educational material, to avoid the identity of educational and control tests. A common approach is random selection of test questions from a given list by their conditional numbers. The questions database should have a large size: the total number of questions in the database should significantly exceed the number of questions in a single test. But database size does not guarantee that the control test will not include questions that have already been used at the stages of training testing or the thematic control test. In addition, this approach does not take into account the logical relationship of questions, which may affect the reliability of knowledge assessment. To solve this problem, use of associative rule search algorithms at the stage of selecting test tasks and including them in the control test was proposed. Associative methods allow identifying the frequency of occurrence of particular questions, to discard the most frequently used ones, to offer options for choosing subsequent questions, taking into account their thematic relationship. The relative newness of research is the use of a modification of associative rule search algorithms in the formation of control test tasks. Practical use of the proposal allows increasing the completeness of the coverage of educational material, objectively evaluating student knowledge level.