Sigrid Ewert

Regular expressions (REs) are often taught to undergraduate computer science majors in the Formal Languages and Automata (FLA) course; they are widely used to implement different software functionalities such as search mechanisms and data validation in diverse fields. Despite their importance, the difficulty of REs has been asserted many times in the literature. Due to their abstract and theoretical nature, students sometimes find REs boring and hard to learn. In addition, a review of existing tools that assist students when learning REs shows that the feedback provided is limited. Therefore, this research set out to automatically generate feedback that contains the type and location of the error and hints on how to fix the error for students learning REs. Principles from compiler construction and the theory of computation were used to develop algorithms for error detection. The algorithms detected the presence, type, and location of errors in students’ RE solutions. The error details identified by the algorithms were then formatted as feedback usable to students. The performance of the algorithms was evaluated using a test dataset consisting of 249 incorrect REs, and the accuracy of predicting the error positions in the incorrect REs was 82%. In comparison with other tools that generate feedback for students learning REs, the generated feedback in this research is more robust because it contains a summary of the type of error present and the location of the errors in both the incorrectly represented strings (counterexamples) and REs. The developed prototype can be enhanced into a full-fledged academic tool to support teachers and students learning REs in live classrooms, be it a physical or virtual class.