Research On Automatic Generation Method Of Use Case Diagram Based On Text Diagram Elements And Their Relationship

In software requirements analysis,the use case diagram intuitively reflects the system functions,which is conducive to the visual understanding of requirements;however,it takes a long time to manually design and draw the use case diagram.In order to improve the efficiency of demand analysis,this paper proposes a method of automatically generating use case diagrams based on text diagram elements and their relationships.This method uses methods such as text classification and use case diagram element identification(hereinafter referred to as diagram elements)to extract the relationship between diagram elements and elements required for drawing use case diagrams from the demand text described in natural language,and automatically generate use case diagrams.Based on the above-mentioned use case diagram automatic generation method,this paper further realizes the use case diagram automatic generation software toolkit.This article first proposes a Dependency parsing-Bidirectional Encoder Representations from Transformers(DPBERT)method based on sentence pattern filtering.Only part of the sentences in the software requirements text are related to the use case diagram elements.Therefore,this research will use the demand text classification method based on sentence pattern filtering to improve the accuracy of the automatic generation of the use case diagram by filtering the sentences in the demand text that have nothing to do with the generation of the use case diagram.In this study,a comparative experiment was carried out by using online public demand documents.The experimental results show that the demand text classification method based on sentence filtering effectively filters the sentences in the text that are not related to the use case diagram.Compared with the automatic generation method of use case diagram based on rule matching,The accuracy of automatically generated results for use case diagrams has increased by 34%.Secondly,based on the hypothesis of correspondence between content words,grammatical components,and graph elements in demand text sentences,this paper proposes a method for automatically extracting graph elements and their relationships based on dependency syntax analysis.This research conducted experiments on the public requirements document test set.The experimental results show that using the method proposed in this paper to analyze the use case diagrams generated by structured or semi-structured requirements documents,the accuracy rate can reach more than 82%compared with the use case diagrams drawn manually.The time complexity of the method proposed in this paper is)(2(48)n,where n is the number of sentences in the article.Based on the above research methods,the author of this article designed and implemented an automatic generation tool for use case diagrams based on diagram elements and their relationships,which were applied to the needs analysis process of actual software projects.Using the tools in this article can accurately analyze the semi-structured actual requirements document.The generated use case diagram has an error of less than 15%compared with the use case diagram manually drawn in the actual application.