LabDER - Relational Database Virtual Learning Environment

Authors

Abstract

This paper describes a virtual learning environment for use in introductory database disciplines that aligns with the professor’s teaching plan and aims to automatically evaluate students’ responses to questions, which may be multiple-choice or discursive or may involve entity relationship diagrams (ERDs) or SQL. The main advantage of LabDER over previous automatic evaluation approaches for ERD and SQL is that it accepts multiple responses and encourages students to develop the best solution through semantic feedback based on compiler theories, software engineering metrics and supervised machine learning. This approach considers the distance of the student’s response from the model response and provides semantic feedback via a blend of compiler and various other metrics, while predicting the student’s grade using a machine learning algorithm. A case study was designed to confirm the approach and 15,158 students’ responses were automatically evaluated. As a result, semantic feedback provided student self-learning through suggestions on the database concepts involved in each solution, which generated a considerable increase in student participation as well as an increase in their average grades. In future work, we will investigate how to include other database topics that can be automatically evaluated, such as query performance and relational algebra.

Keywords

Virtual Learning Environment, Automatic Evaluation, Entity Rela-tionship Diagram, SQL, Compiler, Metrics, Supervised Machine Learning

Conference

38th International Conference on Conceptual Modeling (ER 2019), October 2019

PDF File