Non-Functional Requirement Extracting Methods for AI-based Systems: A Survey

Artificial intelligence (AI) systems are growing rapidly in a variety of fields and have become a part of everyday human life. Traditional software development methods, especially in the requirements engineering domain, are not very efficient in the development of AI-based Systems. Given that the elicitation and analysis of software project requirements is the key to producing a high-quality and efficient product, there must be new approaches to identifying and extracting requirements in the field of AI-based systems. This is done in the research area of Requirements Engineering for Artificial Intelligence (RE4AI). Non-functional requirements are defined in accordance with the system's objectives and operational requirements. They specify the behaviors, limitations, and overall quality of the system; for this reason, they are very important for the success of the system, and lack of attention or neglect to accurately identify it will increase the probability of project failure. In this Paper, we aim to provide a comprehensive overview of the approaches and techniques for extracting non-functional requirements for AI-based systems by reviewing recent papers in this field. The review of these studies shows that there are currently no accurate methods for identifying and extracting non-functional requirements of AI-based Systems, but by relying on traditional methods and redefining their concepts, most of these methods can also be used in this domain. Additionally, new requirements for the field of AI have been introduced that are of high importance.