نوع المشرف

مشرف رئيسي

تاريخ الاشراف على الرسالة من

2015

الى

2017

اسم الطالب

Heba Nassar

ملخص الرسالة

The sequence diagram is a method used for presenting the details of interactions between users and system’s components. A sequence diagram helps in the transition to a more formal level of refinement of the requirement description. Typically, system analysts are responsible for performing this process, and they usually perform the development of sequence diagrams manually. The aim of this thesis is to develop a software tool that generates sequence diagram(s) from the flow of events found in a use case model, which is called an Intelligent Sequence Diagram Generator (ISDG). This tool belongs to Intelligent Computer Aided Software Engineering (I-CASE) tools that have some sort of intelligence to perform those tasks that need human intellectual skills. The thematic role principle is used to distinguish the components of a sequence diagram from the statement of the flow of events. Semantic Role Labelling software type of Natural Language Processing (NLP) tools is used for automatically discovering the thematic role of each word in the input statement. The proposed solution delivered by this research passes through defining a new algorithmic approach for developing sequence diagrams with two versions: manual and semi-automated. The last step is to convert the semi-automated version to a full one by using of Genetic Algorithm (GA) approach for selecting the classification rules of extracting elements of sequence diagram from the natural language form statements of the flow of events. This tool had been implemented using the C# programming language of visual studio that supports embedding other software components and has graphical facilities for drawing the sequence diagram and developing GUI for the tool. The evaluation of the results has been handled using a confusion matrix, in which the accuracy of the ISDG reaches> 77%. Keywords: Software Requirements, UML, Sequence Diagram, Natural Language Processing, Artificial Intelligence.