Unit Testing Maths: Automated Assessment of Mathematic Exercises

摘要

Automated assessment systems for the field of mathematics need to tackle a number of challenging problems. Existing systems focus on the processing of mathematic expressions and their different representations (e.g. x~(1/2) vs.x~(1/2)). In order to correctly assess a student's solution approach, a system has to check intermediate steps and conclusions. Typically, this assessment fails completely in the presence of a small oversight error at the beginning of an exercise. In this paper, we will show, how the idea of unit testing from the field of computer science may be used to detect and handle follow up and cascading errors in mathematic exercises. Thus, resulting in a fairer grading outcome, which is closer to manual grading. The developed technique may be used for large scale (online) courses or acceptance tests.