A Framework to Support Non-Functional Requirements Knowledge Reuse

点击翻译按钮获取中文摘要

Requirements engineers have to address both functional and non-functional requirements (NFR) to develop software systems. Functional requirements are responsible for representing what the system is capable of in terms of available features, while non-functional requirements are known to represent quality attributes. Some common NFRs include privacy, performance, usability, availability and other similar aspects related to the quality of a software system. The proposed research aims at providing an environment that will facilitate a software development process that will favor the reuse of knowledge on satisficing NFRs associated with functional requirements. The first challenge for eliciting NFRs lies on the fact that they are fuzzy in nature and are often missed by both software engineers and stakeholders. Furthermore, choosing one solution to implement one NFR might bring synergies and, perhaps most important, conflicts to other NFRs. Some works have proposed the use of catalogues for capturing knowledge on alternatives to achieve an NFR as a way of helping not only to elicit NFRs but also to reason about the complexity involved in choosing among alternatives. These catalogues are implemented using Softgoal Interdependency Graphs (SIG). SIGs catalogues promote a graphical representation of essential quality characteristics for satisficing a given non-functional requirement. In fact, empirical works have suggested that the use of catalogues can contribute to avoiding omissions and missed conflicts, despite the fact that SIGs do not scale too well. I have recently proposed to use an ontology for NFR supported by semantic web techniques for storing and retrieving alternative solutions for one or more NFR, allowing software engineers to visualise the impacts of each solution. A prototype tool to support this activity was recently finished, and it is prepared to be available in a cloud environment to facilitate a broad use of it in the future. My long-term goal is to engineer the Non-Functional and Design Rationale (NDR) framework to help software engineers to elicit and model non-functional requirements empowered by a knowledge that has previously been elicited. I believe that the use of a well-defined knowledge base can play a major role to achieve this goal. Therefore, the NDR Framework will emerge as the result of further developing our ontology and create methods, techniques and tools to store and retrieve knowledge on implementing non-functional requirements. The framework will also provide methods to integrate NFRs into functional specifications. The NDR Framework will be first made available to the academic community to evaluate it and to later contribute to the knowledge repository. After the framework is stable and mature, we will also allow members from the industry to query the knowledge base and, in the long run, to submit contributions to be added to the knowledge base.