An approach to support the requirements specification in agile software development

Abstract

Although Agile Software Development (ASD) has grown in recent years, research evidence points out several limitations concerning its requirements engineering activities. It was observed that an inadequate specification acts as a catalyst to others problems, such as low productivity of the team and difficulty in maintaining software. Improving the quality of Software Requirements Specifications (SRS) may help gaining a competitive advantage in the software industry. The goal of this study is to investigate the phenomenon of the requirements specification activity in ASD, discuss relevant findings of this phenomenon to industrial practice, and propose practices to write a SRS targeted to development team. First, a Systematic Mapping (SM) study was conducted to characterize the landscape of requirements engineering in ASD. The thematic synthesis method was used to code and synthesize the data collected from the primary studies selected. After that, some of the challenges pointed out in the SM were investigated in more depth in six industrial case studies. Data collected from documents, observations, and interviews with software engineers were triangulated, analyzed, and synthesized using techniques of grounded theory and metaethnography. The analysis and cross-synthesis of the case studies resulted in a model that defines the simplicity and objectivity as essential quality factors of SRS in ASD. The main factors that affect the quality are related to the customer-driven nature that tends to leave the prolix SRS, hindering the understanding of the software engineers, as they are, at the same time, insufficient to support coding, testing and maintenance tasks. One approach was proposed to provide a SRS closer to the development needs, addressing some of the quality factors of the model. Empirical studies that evaluated the approach show that the design practices used in the proposed approach have the potential to reduce the gap between the problem and the solution domains, producing an objective SRS that is team-driven and closer to that will be implemented.