Answer Generation for Set Type Math Word Problems

In this paper, an automatic math word problem solver is presented for GCE Ordinary Level set theory related problems that use Venn diagrams. In order to model all basic set relationships and operations required to solve the problems, a data representation is introduced that uses binary representations. After categorizing set problems in O/L Mathematics examinations including GCE, IGCSE, GCSE, and JCE, two types of problems are selected to solve using this representation. The solver is implemented using regular expressions and POS tag based information extraction to filter relevant data from a given problem text. Then this information is mapped into the above data representation in order to derive equations that solve the problem. On the end-to-end evaluation, an accuracy of over 90% is achieved for the selected set of math word problem types.