Towards an Automated Assistant for Generating Mathematics Problems

ABSTRACT

High school learners in low-income countries are negatively impacted by the SARS Covid-19 pandemic. Consequently, university lecturers of mathematics will have to offer remedial classes to bridge the gap. Since they cannot create practice problems at scale for each student's needs, there is a need for computational tools to do so. There are no existing and published formalisations of mathematical problems that abide by the South African curriculum to allow the automatic generation of problems. We aim to address this gap by formalising exam problems written by grade 12 South African learners in the period 2008–2020. We evaluate the problem types by demonstrating 65% coverage of the 74 matric rewrite problems from the years 2011–2018. The presented problem formalisations allow the generation of maths problems to be used for student-led remedial practice. © 2022, Springer Nature Switzerland AG.