Exploring a Comprehensive Approach for the Automated Assessment of UML

Abstract

UML is an important tool in structured software design and is commonly taught in undergraduate software courses. UML defines a complex set of linked notations and mastery requires instruction and examples over many taught courses. Such examples are typically disparate, modelling subsets of distinct systems. Teaching UML benefits from an end-to-end approach where consistency between modelling, design and implementation are equally emphasised and regular feedback is provided. A drawback of this is that assessing learner-derived UML models is a time-intensive and error-prone task if both formative and summative assessment is required. In this paper a novel framework is presented for the automated assessment of UML. The framework allows learners to be provided with automatically generated formative feedback for self-directed learning in the development of UML skills. Emphasis is placed upon the identification of the consistency and coverage of learner diagrams as this is an important skill in the application of UML. By integrating an implementation of this framework, instructor-based UML teaching can be supplemented with an end-to-end tool which allows learners to receive automated formative feedback in their understanding and usage of UML.