Program Repair for Intelligent Tutoring and Programming Education
Abstract:
Automated program repair is a technology for automated rectification of errors and vulnerabilities in programs. This technology can be used for intelligent tutoring of programming – where student assignments are compared with reference assignments to compute feedback to students. In this thesis, we report on the design, implementation, real-life evaluation, as well as the experience of hundreds of students, for such an intelligent tutoring system. We also show how the core program repair technology can combined with grading rubrics to provide automated grading of programming assignments. Such a grading goes beyond test-suite based grading since the aim is to find concepts which are not understood by students. The intelligent tutoring system has been conducted as a real-life project in the course of teaching, and has been used by tutors / students of a first year programming course at National University of Singapore. In the last part of the talk, we report the experience of using the intelligent tutoring system in conjunction with Large Language Models (LLMs) on data structure and algorithms tasks to provide a forward-looking perspective about the teaching of other Compunter Science courses in future.