Enhancing fault-detection in Conceptual Modelling: A case study of deadlock identification and detection

In the ever-evolving world of software development, conceptual modelling (CM) has gained significant prominence, especially with the advent of low-code/no-code development techniques. Rooted in model-driven development (MDD), these approaches aim to simplify coding complexity and bridge the gap between business and IT. However, challenges remain, particularly when it comes to identifying and rectifying errors in models, which can lead to unreliable and error-prone software.

One critical software defect that modellers frequently face is the potential for deadlock situations. These occur when a system’s components become obstructed and are unable to complete assigned tasks. Our recent study delved into how an Automated Feedback System (AFS) can assist novice modellers in identifying and correcting possible deadlock situations in conceptual models.

The study employed a two-group posttest-only experimental design, aimed at assessing the impact of an AFS-enhanced approach on novice modellers’ ability to detect and rectify deadlock-related errors within the context of a CM course. Despite our best efforts, the experiment did not conclusively demonstrate a significant improvement in performance with the AFS-enhanced approach.

However, we didn’t stop there. By redividing the participants based on their self-reported usage of a model-simulation tool that provides feedback on the model coverage of their testing efforts, we found evidence that the initial negative result was partially due to an imbalanced experimental group division.

By redividing the participants based on their self-reported usage of a model-simulation tool during the experiment, we found evidence supporting one of our hypotheses.

The study concludes with several proposed improvements to the experimental design and discusses the implications for teaching of fault-detection skills in conceptual modelling  based on our findings. These insights pave the way for more effective and reliable CM practices in the future.

For more details about our research, you can read the full paper at: 🔗 https://lnkd.in/dxRTVX5s