Software traceability serves a critical role in ensuring that software systems operate correctly. It is used to support a wide variety of software engineering activities such as change management, compliance verification, and safety analysis. Unfortunately current practices fall far short of delivering cost-effective traceability, primarily because creating and managing trace links in large and/or complex systems is time-consuming, arduous, and error-prone. These problems were highlighted in a recent report entitled 'Critical Code: Software Producibility for Defense' commissioned by the Department of Defense. The report stressed the need for the research community to develop cost-effective and accurate traceability solutions. While state of the art tracing techniques offer significant promise for reducing the cost and effort of tracing, they fall short of meeting industrial needs primarily because the quality of the generated links is imprecise. This work will investigate ways to integrate techniques from feature modeling, product line development, artificial intelligence and machine learning to deliver a dynamically configurable trace infrastructure. The framework will then be used to investigate and integrate a broad set of novel tracing techniques which are expected to significantly improve the quality of generated trace links.
The results of the project will contribute towards the development of software intensive systems, especially safety-critical ones in which traceability is mandatory. Technology transfer will be facilitated by delivering solutions on the TraceLab platform, disseminated via the Center of Excellence for Software Traceability (CoEST.org) and through training materials targeted at industrial users. Ongoing research opportunities will be provided for a diverse group of undergraduate and graduate students, and pedagogical materials will be developed and made publicly available for use in a variety of courses on requirements engineering, software engineering and software architecture.
