Automated Generation of Computer Graded Unit Testing-Based ProgrammingAssessments for Education

摘要

Automatic assessment of code, in particular to support education, is an important featureincluded in several Learning Management Systems (LMS), at least to some extent. Several kindsof assessments can be designed, such as exercises asking to “fill the following code”, “write afunction that”, or “correct the bug in the following program”, for example. One difficulty forinstructors is to create such programming exercises, in particular when they are somewhatcomplex. Indeed, instructors need to write the statement of the exercise, think about the solutionand provide all the additional information necessary to the platform to grade the assessment.Another difficulty occurs when instructors want to use their exercises on another LMS platform.Since there is no standard way to define and describe a coding exercise yet, instructors have tore-encode their exercises into the other LMS. This paper presents a tool that can automaticallygenerate programming exercises, from one single and unique description, and that can besolved in several programming languages. The generated exercises can be automaticallygraded by the same platform, providing intelligent feedback to its users to support theirlearning. This paper focuses on and details unit testing-based exercises and provides insightsinto new kinds of exercises that could be generated by the platform in the future, with someadditional developments.