Method for Adaptive Semantic Testing of Educational Materials Level of Knowledge

Abstract

The article considers the method that allows to calculate the estimation of knowledge level of educational materials by using indicators of semantic importance of key terms for adaptive selection of test tasks in testing. Each test task allows to purposefully check the level of knowledge of separate semantic units of educational materials -semantic terms (words, phrases). It is assumed that increasing the depth of learning of educational material semantic content has the effect of learning less semantically important units of educational materials. Semantic terms are related to the semantic structure of educational material in the form of rubricational system. Test tasks are related to fragments of educational material content, the knowledge level of which they test. In the process of testing, the level of knowledge of each of semantic structure elements of the educational material is consistently adaptively determined, and the final grade is calculated based on these results. As a result of adaptive selection of test tasks, their maximum possible diversity in the final set is reached, because following criteria are considered: test task has not been used yet, relevant fragment of educational content has not been checked, test tasks of corresponding type were used the least, the test task does not contain less important semantic units. The developed method of adaptive semantic testing of knowledge level of educational materials makes possible to use different algorithms for starting testing (regressive, progressive, medianic, etc.) and different algorithms for knowledge level estimation (average, absolute limit, etc.). Applied investigations of the effectiveness of the developed method in comparison with the traditional algorithm for selecting test tasks established, that testing speed increased an average of 20.53% faster test, and to determine the level of knowledge required the use of an average of 19.33% fewer test tasks.