BERT Goes to SQL School: Improving Automatic Grading of SQL Statements

Automated grading of SQL queries is a challenging task due to the complexity of the language and the variety of acceptable solutions for a given problem. In this paper, we propose a novel approach that leverages deep learning with a BERT model to understand the syntax and semantics of SQL statements. By training BERT on a dataset of SQL queries and their corresponding grades, we create a model that can automatically grade new questions accurately. Our experiments demonstrate that the proposed methodology achieves high accuracy and consistency in grading SQL queries, outperforming existing state-of-the-art models. Furthermore, we provide an analysis of the model's explainability, revealing a new capability that can be extremely beneficial for understanding the decision-making process. Overall, our work demonstrates the potential of deep learning with BERT for improving the efficiency and accuracy of SQL query grading.