Enhancing UML class diagram abstraction with knowledge graph

Abstract

Model-Driven Engineering (MDE) alleviates the cognitive complexity and effort spent on software development by generating codes from models. In MDE, models should be accurate, refined, reliable and efficient. Class diagram is a structural abstraction of a real system and usually used in software design. A better designed class diagram could lead to a better system. In this paper, we proposed a knowledge graph based method to improve class diagrams. We took knowledge graph as the media layer for easier information introduction, and proposed methods to map data, information and knowledge between class diagrams and knowledge graphs bidirectionally. Based on the added knowledge source, we designed hierarchical clustering algorithm to abstract the class diagram, and finally we generated abstracted class diagrams automatically.