Contributions to management and validation of non-functional requirements

Sammanfattning: Non-functional requirements (NFRs) are essential when considering software quality in that they shall represent the right quality of the intended software. It is generally hard to get hold of NFRs and to specify them in measurable terms, and most software development methods applied today focus on functional requirements (FRs). Moreover, NFRs are relatively unexplored in the literature and knowledge regarding real-world treatment of NFRs is particularly rare.A case study and a literature survey were performed to provide this kind of knowledge, which also served as a problem inventory to outline future research activities. An interview series with practitioners at two large software development organizations was carried out. As a major result, it was established that too few NFRs are considered in development and that they are stated in vague terms. Moreover, it was observed that organizational power structures strongly influence the quality of the forthcoming software, and that processes need to be well suited for dealing with NFRs.It was selected among several options to explore how processes can be better suited to handle NFRs by adding the information of actual feature use. A case study was performed in which the feature use of an interactive product management tool was measured indirectly from log files of an industrial user, and the approach was also applied to the problem of requirements selection. The results showed that the idea is feasible and that quality aspects can be effectively addressed by considering actual feature use.An agenda for continued research comprises: further studies in system usage data acquisition, modelling of NFRs, and comparing means for predicting feasibility of NFRs. One strong candidate is weaving high-level requirement models with models of available components.