QAExtractor: A Quality Attributes Extraction Framework in Agile-Based Software Development

Abstract

Many software projects fail due to an inadequate understanding of their quality at the initial stages of the development.Ahmed, Mohsin The quality of the software heavily depends on non-functional requirements.Khan, Saif Ur Rehman But requirements elicitation team pay less attention towards non-functional requirements, during the requirements elicitation stage of software development life cycle.Alam, Khubaib Amjad In agile-based software development, the requirements elicitation team only focuses on the user stories (i.e. functional requirements in an agile-based software development context). The non-functional requirements are neglected, and team members are left only with the user stories. This makes it very difficult for them to make decisions about quality while holding user stories in their hands. Furthermore, it is essential to affirm the quality of the software as early as possible. This is mainly because the quality of software can negatively affect the different artefacts of the software at later development stages, i.e. designing, developing, etc. The early stage conformance of software quality is more important in the Agile-based Software Development context, where the requirements are more volatile than any other development environments. Hence, there is a need to develop such an automated framework which can extract quality attributes from user stories automatically. In this work, we propose a Quality Attributes Extraction Framework, named as QAExtractor, for the extraction of key quality attributes from functional requirements (i.e. user stories in Agile-based Software Development context). The core of this framework is based on Natural Language Processing. The proposed technique grounds on the regular expressions, which generalise the user stories for a specific quality factor and attribute. The quality factor defines the context of the user story, i.e. security, while quality attribute states the quality aspect, i.e. integrity. The effectiveness of the proposed technique is validated using a case study. The experimentation results revealed that the proposed framework outperforms the existing ones in terms of accuracy, precision, recall and F measure.