Abstract:

The use of Information Communication Technology (ICT) tools in education is almost commonplace. The COVID19 epidemic has, wittingly or unwittingly, forced education to the online space. This means that in addition to the delivery of course materials, testing and teaching was done electronically. The focus of our project was on checking texts.

The aim of the study is to facilitate the improvement of the increasingly common electronic tests, thus speeding up and facilitating the work of trainers. Systems designed for this purpose cannot only be used to improve shorter multi-sentence answers. They can also be a quick and easy way to interpret lengthy essays or graduation tests. In addition to describing research in the related field, the paper also presents the development of an own project. Our result is a system based on neural network architecture, which is able to analyse authentic sentences written in its own words and classify them into a set of correct and incorrect answers. Classification is based on large quantities and high-quality samples, which are essential for optimal operation. Focusing on mathematics, we worked with upper primary school mathematics concepts. During the preparation phase we relied on definitions published in literature dealing with the specified topics. In addition, we processed the answers which were built up entirely from the students’ vocabulary who took the tests. Using the extracted data sample, we carried out the live testing on other students. Our results suggest that similar procedures can significantly simplify the evaluation of electronic tests.

In this paper, we present the architecture and operating principles of this system. Practical results will be presented to illustrate the potential uses of the application. Lastly conclusions are drawn from the evaluation of the obtained outputs. Finally, based on the results obtained, we outline the concept of further development options.

Keywords:

Application, software, educational tool, testing, digitalisation, neural network, computer software on education.