Software Engineering Events
|13.04.2021 16:00 SE Colloquium https://uol.de/se?vc-oberseminar|
Johannes Martin Legler:
Automated Evaluation of Modeling Tasks for Self-study
The University of Oldenburg teaches programming, modeling, and other theoretical contents of Computer Science. It has been observed that the students often seem to struggle with their study, due to the amount of lecture content, the limited time, and the needed self-study at large. Self-study for modeling is especially difficult for them, because of the solution ambiguity for each task, where they need to consider alternative (completely different) solutions, as well as variants of solutions, which are only marginally different, when evaluating their own solution.
To support the students with their self-study on modeling tasks, this thesis proposes a concept called ÆvaMoT (Automatic Evaluation of Modeling Tasks), which allows for automatic evaluation of student solutions with appropriate feedback. For that it answers the questions on how sample solutions for these tasks can be represented to be efficiently and automatically analyzable and how the results of that can be used for fully automatic evaluation and annotation of student solutions.
The proposed concept consists firstly of an UML profile containing UML stereotypes, which can be attached to any element UML diagram element. With them, instructors can provide feedback templates based on which the tool can generate customizable feedback messages to the students. Secondly, a GReQL query graph is constructed from all available sample solutions for a task, with which the student solution is ultimately evaluated.