Towards Intelligent Requirements Engineering: A Systematic Review of AI-Enabled Solutions

Abstract

Requirements Engineering (RE) is an important stage during software development which aims to explore, analyze, validate, prioritize and manage the requirements of stakeholders prior to the system implementation. Traditional RE approaches suffer from problems of natural-language ambiguity, evolving needs, extensive documentation, conflicting views among stakeholders, and low levels of automation. This Systematic Literature Review (SLR) examines 43 studies to gain an understanding of the role of Artificial Intelligence (AI) in improving RE activities. AI technologies such as Natural Language Processing (NLP), Machine Learning (ML), Large Language Models (LLMs), Generative AI (GenAI), prompt engineering, knowledge graphs, ontologies, optimization algorithms, and multi-agent systems are increasingly being used, as highlighted in the review. AI is being broadly utilized in the requirements elicitation, analysis, specification, validation, prioritization, traceability, and semantic management. NLP and ML are integral to text analysis and classification, and LLMs and GenAI are vital to automated requirement generation, documentation, and validation. Intelligent and integrated RE frameworks, such as knowledge-graph approaches, ontology-based models and automated multi-agent systems are emerging trends. There are, however, problems like hallucination, explainability, sensitivity to prompts, privacy issues, ethical concerns, poor reproducibility, and lack of validation in industry that are still a barrier to adoption. The study found that while the use of AI in RE is promising, it needs expert oversight, thorough testing, and practical solutions to be ready for the market.