Performance evaluation of a new approach for automatic question production

Abstract

Nowadays, е – testing is an often used method for evaluation in the process of learning. In this paper, we discuss the е – testing problem of creating large question set that will reflect the knowledge of some domain. A new model of E – testing is introduced with a proposal of a new solution to the problem of creation a large question set for a given domain. Then, we present a methodology for comparison of the results and the contribution of the new model and realization on the automated creation of large number of questions, and we evaluate the quality and the vulnerability of the question set, as well. It is shown that the new model increases the speed of question production by more 10 times.