| Abstract [eng] |
Software development starts with business analysis. Business clients usually formulates their requirements using natural language. There is a need to automate, as far as possible, creation of preliminary UML models. The aim of this project is to analyze semi-automated and automated methods and tools of developing preliminary UML models from natural language. The tool would help to analyze primary requirements, given in the form of natural language and create preliminary UML models. This work consists of several main stages: Preparation Design Realization Testing Experiment This paper describes a tool that helps to create preliminary UML models from requirements written in natural language. The main advantage of this tool is that it has web interface, and there is no need to install any additional software. Analyst, using his network browser, enters online tool, where he can paste his requirements. After pasting, requirements are being saved to database for further processing. Analyst can now analyze requirements and assign elements, such as Actors and Use cases, to words or phrases. If use case has included or extended use cases, analyst can assign them too. Finally, as a result, analyst can export XMI model, which can later be imported to CASE tool. After implementation, an experiment was conducted. The tool was evaluated by precision and recall criteria. The results shows that tool for creating preliminary UML models from text meets precision and recall criteria by 100% when requirements are written in highly structured text. |
