Supporting Classification of Software Requirements system Using Intelligent Technologies Algorithms

Abstract

The important first stage in the life cycle of a program is gathering and analysing requirements for creating or developing a system. The classification of program needs is a crucial step that will be used later in the design and implementation phases. The classification process may be done manually, which takes a lot of time, effort, and money, or it can be done automatically using intelligent approaches, which takes a lot less time, effort, and money. Building a system that supports the needs classification process automatically is a crucial part of software development. The goal of this research is to look into the many automatic classification approaches that are currently available. To assist researchers and software developers in selecting the suitable requirement categorization approach, those requirements were divided into functional and non-functional requirements. since natural language is full of ambiguity and is not well defined, and has no regular structure, it is considered somewhat variable. This paper presents machine requirement classification where system development requirements are categorized into functional and non-functional requirements by using two machine learning approaches. During this research paper, MATLAB 2020a was used, as well as the study's results indicate When applying Multinomial Naive Bayes technology, the model achieves the highest accuracy of 95.55 %,93.09 % sensitivity, and 96.48 % precision, However, when using Logist Regression, the suggested model has a classification accuracy of 91.23 %,91.54 % sensitivity, and 94.32 % precision.