Toward collisions produced in requirements rankings: A qualitative approach and experimental study

Abstract

Requirements prioritization is an important issue that determines the way requirements are selected and processed in software projects. There already exist specific methods to classify and prioritize requirements, most of them based on quantitative measures. However, most of existing approaches do not consider collisions, which are an important concern in large-scale requirements sets and, more specifically, in agile development processes where requirements have to be uniquely selected for each software increment. In this paper, we propose QMPSR (Qualitative Method for Prioritizing Software Requirements), an approach that features the prioritization of requirements by considering qualitative elements that are related to the project's priorities. Our approach highlights a prioritization method that has proven to reduce collisions in software requirements rankings. Furthermore, QMPSR improves accuracy in classification when facing large-scale requirements sets, featuring no scalability problems as the number of requirements increases. We formally introduce QMPSR and then define prioritization effort and collision metrics to carry out comprehensive experiments involving different sets of requirements, comparing our approach with well-known existing prioritization methods. The experiments have provided satisfactory results, overcoming existing approaches and ensuring scalability.