Automatic Generation of UML Class Diagrams for Object-oriented Design and Programing Course

UML Class diagram generation from textual requirements is an important task in object-oriented design and programing course. This study proposes a method for automatically generating class diagrams from Chinese textual requirements on the basis of Natural Language Processing(NLP) and mapping rules for sentence pattern matching. First, classes are identified through entity recognition rules and candidate class pruning rules using NLP from requirements.Second, class attributes and relationships between classes are extracted using mapping rules for sentence pattern matching on the basis of NLP. Third, we developed an assistant tool integrated into a precision micro classroom system for automatic generation of class diagram, to effectively assist the teaching of object-oriented design and programing course. Results are evaluated with precision, accuracy and recall from eight requirements of object-oriented design and programing course using truth values created by teachers. Our research should benefit beginners of objectoriented design and programing course, who may be students or software developers. It helps them to create correct domain models represented in the UML class diagram.