Development of an Automated Assessment Tool "Silmused" in the Databases course

Abstract

This thesis presents the development and evaluation of Silmused, an automated assessment tool designed for the Databases LTAT.03.004 course at the University of Tartu. The tool addresses the increasing workload of teaching staff and the need for timely feedback for students by automating the evaluation of database assignments. Built as a Python library, Silmused integrates with the Lahendus platform to provide instant and consistent feedback to enhance the assessment process. The thesis covers the evaluation of Silmused through questionnaires administered to both teaching staff and students. Results indicate significant time savings for teaching staff, with grading time reduced from 40–100 minutes per task to approximately 20 minutes per task. Students reported high satisfaction with the clarity and immediacy of feedback, though some noted challenges with complex error messages and minor syntax variations. Key contributions include the tool’s modular design for compatibility and its alignment with formative learning principles. While Silmused has proven effective in enhancing efficiency in teaching and learning outcomes, future improvements could focus on refining feedback clarity, accommodating syntax and flexibility