Reducing Cognitive Load in Learning to Model UML Sequence Diagrams

This paper demonstrates how cognitive load theory can be used to improve learning outcomes by presenting a tool capable of assisting novices to learn to model sequence diagrams effectively. Learning sequence diagrams is known to lead to heavy cognitive load as they must be consistent with the class diagram, while discharging all the responsibilities specified in the underlying use case. Moreover, novices must also consider the various design options and their impact on the qualitative aspects of the model. Our tool allows cognitive load to be better managed by using a ‘divide and conquer’ approach. In the initial stage, only consistency with the class diagram is enforced. In the second stage, invalid messages are disallowed by tracking the active entity and the current knowledge state in entities. In the third stage, students will not be allowed to submit a diagram until the stated use case goals are met. In the final stage, qualitative feedback and marks are awarded based on established metrics, and students are allowed to improve their scores by resubmitting the model. Students’ performance and feedback on the tool show that our novel framework combining scaffolding with a form of gamification has helped to improve the learning outcomes in modelling substantially, especially among stragglers. One benefit of our approach is that it can be adapted to other areas where students may be cognitively challenged.