From natural language requirements to UML class diagrams

Unified Modeling Language (UML) is the most popular modeling language for analysis, design and development of the software system. There has been a lot of research interest in generating these UML models, especially class diagrams, automatically from Natural Language requirements. The interest in class diagrams can be attributed to the fact that classes represent the abstractions present in the system to be developed. However, automated generation of UML class diagrams is a challenging task as it involves lot of pre-processing or manual intervention at times. In this paper, we present dependency analysis based approach to derive UML class diagrams automatically from Natural Language requirements. We transform the requirements statements to an intermediary frame-based structured representation using dependency analysis of requirements statements and the Grammatical Knowledge Patterns. The knowledge stored in the frame-based structured representation is used to derive class diagrams using rule-based algorithm. Our approach has generated similar class diagrams as reported in earlier works based on linguistic analysis with either annotation or manual intervention. We present the effectiveness of our approach in terms of recall and precision for the case-studies presented in earlier works.