Date of Award
Doctor of Philosophy in Information Systems (DISS)
College of Engineering and Computing
Francisco J. Mitropoulos
agile software development, NFR priority, non functional requirements, optical character recognition, software engineering, software quality, Computer science
Agile software engineering has been a popular methodology to develop software rapidly and efficiently. However, the Agile methodology often favors Functional Requirements (FRs) due to the nature of agile software development, and strongly neglects Non-Functional Requirements (NFRs). Neglecting NFRs has negative impacts on software products that have resulted in poor quality and higher cost to fix problems in later stages of software development.
This research developed the CEP “Capture Elicit Prioritize” methodology to effectively gather NFRs metadata from software requirement artifacts such as documents and images. Artifact included the Optical Character Recognition (OCR) artifact which gathered metadata from images. The other artifacts included: Database Artifact, NFR Locator Plus, NFR Priority Artifact, and Visualization Artifact. The NFRs metadata gathered reduced false positives to include NFRs in the early stages of software requirements gathering along with FRs. Furthermore, NFRs were prioritized using existing FRs methodologies which are important to stakeholders as well as software engineers in delivering quality software. This research built on prior studies by specifically focusing on NFRs during the early stages of agile software development.
Validation of the CEP methodology was accomplished by using the 26 requirements of the European Union (EU) eProcurement System. The NORMAP methodology was used as a baseline. In addition, the NERV methodology baseline results were used for comparison. The research results show that the CEP methodology successfully identified NFRs in 56 out of 57 requirement sentences that contained NFRs compared to 50 of the baseline and 55 of the NERV methodology. The results showed that the CEP methodology was successful in eliciting 98.24% of the baseline compared to the NORMAP methodology of 87.71%. This represents an improvement of 10.53% compared to the baseline results. of The NERV methodology result was 96.49% which represents an improvement of 1.75% for CEP. The CEP methodology successfully elicited 86 out of 88 NFR compared to the baseline NORMAP methodology of 75 and NERV methodology of 82. The NFR count elicitation success for the CEP methodology was 97.73 % compared to NORMAP methodology of 85.24 %which is an improvement of 12.49%. Comparison to the NERV methodology of 93.18%, CEP has an improvement of 4.55%. CEP methodology utilized the associated NFR Metadata (NFRM)/Figures/images and linked them to the related requirements to improve over the NORMAP and NERV methodologies. There were 29 baseline NFRs that were found in the associated Figures/images (NFRM) and 129 NFRs were both in the requirement sentence and the associated Figure/images (NFRM).
Another goal of this study was to improve the prioritization of NFRs compared to prior studies. This research provided effective techniques to prioritize NFRs during the early stages of agile software development and the impacts that NFRs have on the software development process. The CEP methodology effectively prioritized NFRs by utilizing the αβγ-framework in a similarly way to FRs. The sub-process of the αβγ-framework was modified in a way that provided a very attractive feature to agile team members. Modification allowed the replacement of parts of the αβγ-framework to suit the team’s specific needs in prioritizing NFRs. The top five requirements based on NFR prioritization were the following: 12.3, 24.5, 15.3, 7.5, and 7.1. The prioritization of NFRs fit the agile software development cycle and allows agile developers and members to plan accordingly to accommodate time and budget constraints.
Richard Rabin Maiti. 2016. Capturing, Eliciting, and Prioritizing (CEP) Non-Functional Requirements Metadata during the Early Stages of Agile Software Development. Doctoral dissertation. Nova Southeastern University. Retrieved from NSUWorks, College of Engineering and Computing. (968)