Arabic User Requirements Classification Using Machine Learning

• Published in: International Conference on Information Technology Research (Online) pp. 483 - 488
• Main Authors: Amro, Reem, Althunibat, Ahmad, Hawashin, Bilal
• Format: Conference Proceeding
• Language: English
• Published: IEEE 09.08.2023
• Subjects: Costs; Functional Requirement; Information technology; Machine learning; Manuals; Natural languages; Non-Functional Requirement; Requirements Classification; Requirements engineering; Software; Software Engineering
• ISSN: 2831-3399
• DOI: 10.1109/ICIT58056.2023.10225936

Requirement engineering is a crucial step in software engineering as it forms the foundation for all subsequent stages and significantly affects whether software development is successful or unsuccessful. The same software requirement document generally includes both functional and non-functional requirements are essential. However, expressing these requirements in natural language necessitates considerable human effort to classify them. Manual classification can be laborious, time-consuming, expensive, and prone to inaccuracies. Inaccurate classification can result in misunderstandings or ambiguities in requirements, leading to incomplete products that fail to meet customer demands. While many studies have proposed English language requirements classification techniques, there is an absence of research in Arabic requirements classification. Moreover, there aren't many Arabic datasets that are openly accessible in this sector. In this paper, we offer a method for gathering Arabic data sets for needs and utilizing machine learning techniques to automatically categorize software requirements written in Arabic into functional and non-functional requirements. This study intends to assist software developers in time savings, lower the cost and effort associated with the manual classification process, and improve the effectiveness of the requirements engineering phase. Further gives scholars in this subject a new topic to explore.