A Requirement Analysis Framework Based on Feasibility Analysis, Collateral Analysis, KAOS and Feature Modelling

Abstract

Building the right software starts with understanding and answering key questions i.e. what is truly needed, checking the capability of the team on can they build, how they will build, will they be reimbursed correctly? These questions are summation of the requirement engineering or requirement analysis, an essential step of the software development methodology. As Natural Language Processing (NLP) grows more powerful, particularly with new Artificial Intelligence (AI) like Large Language Models (LLM), getting these requirements right is more necessary than ever. This paper proposes a new framework for requirement analysis formulated on structured methods like Delphi and Nominal Group Technique (NGT) gathering expert input and building a solid work. Our approach works step by step: it first checks a project's basic feasibility, then uses MEASUR’s Collateral Analysis to consider greater impact and risks. Next, it clarifies all aims using a KAOS goal model, and finally maps out all the parts and their connections in a Feature Diagram. A case study at the university of swat was carried out on NLP and LLMs based projects for requirement analysis. The approach was tested against traditional, radical, contemporary and agile techniques for requirement analysis. Our proposed framework performed 12% better than the traditional techniques, 17% better than agile and contemporary techniques and 14% better than Radical technique for Requirement analysis, proving its practical value for creating reliable and well planned NLP applications.