Interactive Pedagogical Agents for Learning Sequence Diagrams

Students struggle to learn sequence diagrams (SDs), as the designs must meet the requirements without violating the constraints imposed by other UML diagrams. Providing manual timely feedback, though effective, cannot scale for large classes. Our pedagogical agent combining data dependencies and quality metrics with rule-based techniques capturing consistency constraints allowed generation of immediate and holistic feedback. The scaffolding approach helped to lower the cognitive overload. The pre- and post-tests and survey results revealed substantially improved learning outcomes and student satisfaction.