Abstract
The increasing demand for programming education has given rise to all kinds of online evaluations, such as Massive Open Online Courses (MOOCs) focused on introductory programming assignments (IPAs), especially over the last few years due to the coronavirus outbreak. As a consequence of a large number of enrolled students, one of the main challenges in these courses is to provide valuable and personalized feedback to students. This personalized feedback can be provided as a list of possible repairs to a student’s program. Typically semantic program repair tools repair an incorrect program using a correct implementation for the same IPA. In order to compare both programs, a relation between both programs’ sets of variables is required. Thus, in this work, we propose to learn how to map the set of variables between different small imperative programs based on both programs’ abstract syntax trees (ASTs) using graph neural networks (GNNs).
Pedro Orvalho
Computer Science Ph.D. Student
My research interests include Automated Reasoning, Program Repair and Program Synthesis.