Designing a Framework for Evaluating Automated Requirements Elicitation Systems in Agile Projects: A Proposal Based on Quality and Reliability Criteria

Résumé fourni par la source

Designing a Framework for Evaluating Automated Requirements Elicitation Systems in Agile Projects: A Proposal Based on Quality and Reliability Criteria Hosein Tanabi1 1- PhD Student, Department of Computer Engineering, Islamic Azad University, Babol Branch, Babol, Iran Abstract Today, agile software development methods are recognized as a dominant approach in the software industry due to their flexibility and rapid response to changes. However, elicitation and management of user requirements in these dynamic environments remain a fundamental challenge [1]. The emergence of automated systems based on Natural Language Processing (NLP) and Large Language Models (LLMs) promises a transformation in this domain [2]. Nevertheless, the systematic evaluation of quality and reliability of these systems, particularly in the context of the Persian language, requires further investigation. This research aims to design a conceptual evaluation framework for assessing the quality and reliability of automated requirements elicitation systems in agile projects. The proposed framework is based on INVEST criteria and the INCOSE standard, and it has the potential to be implemented on real-world datasets of user stories. While some LLM-based systems demonstrate high capability in extracting functional requirements [3], they face challenges in identifying non-functional requirements and trust-related constraints, necessitating algorithmic improvements and the adoption of hybrid approaches. It is hoped that the presented framework can serve as a foundation for future research in the evaluation of automated requirements elicitation systems. Keywords : Automated Requirements Elicitation, Agile Systems, Evaluation Framework, Quality Assessment, Large Language Models, Persian Language Processing

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.