Hamilton, MargaretAlhazmi, Sohail2024-02-292024-02-292024-02-19https://hdl.handle.net/20.500.14154/71529Sequence diagrams, which visually depict the interaction between objects, are critical components of modern software development. Sequence diagrams enable experienced developers to create more effective and sustainable solutions by detecting and avoiding potential issues during the design stage. However, learning sequence diagrams places a heavy cognitive load on students, who must (i) ensure consistency with class diagrams, (ii) check the validity of dispatched messages, (iii) meet the goals stated in use cases, and (iv) justify their design decisions taking into account qualitative attributes. No attempt appears to have been made to produce a pedagogical framework capable of providing ongoing diagnoses and support for students. The aim of this thesis is to find novel ways to engage and support diverse incoming students to learn how to model. The first research task is addressed by the proposed rule-based framework, which is able to give immediate feedback by capturing the interdependencies between class and sequence diagrams. The second research task is addressed by tracking the knowledge state and by incorporating the design by contract technique. The design by contract precondition ensures necessary knowledge elements are present at each entity before a message can be dispatched. The third research task is addressed by ensuring the created and submitted sequence diagrams meet use case goals. This is achieved by capturing the postconditions and by incorporating an Artificial Intelligence (AI) planning technique into the framework. The final research task is addressed by providing qualitative feedback and marks based on qualitative metrics, designed to ensure maintainability, reusability and performance. A gamification strategy allowing multiple submissions with the goal of improving design scores, helped to motivate and engage students. The proposed pedagogical framework was based on a scaffolded approach to reduce the cognitive load on novices by requiring them to focus only on one aspect at each intermediate stage. A mixed methods design including qualitative and quantitative data was used to evaluate the proposed approach. The approach was validated by conducting experiments across different cohorts of students studying software engineering courses. Experimental results were collected through pre-and post-tests, survey results, expert interviews and data recorded by the tool. These sources revealed that the novel pedagogical framework substantially improved learning outcomes. The greatest improvement was noted among stragglers, who were the main target group of this thesis. The proposed framework is suitable for large classes and online teaching because the feedback and marks are automatically generated. The framework can also be adapted to other areas where diverse students face cognitive overload.191eninteraction diagramspedagogical toolscaffolded approachUML modellingThesis