| The effective management of information is a fundamental task in artifact--rich domains such as software engineering and e--Science. In software engineering, numerous heterogeneous artifacts are produced during the software development lifecycle, motivating the need for software traceability. Traceability aims to identify the relationships between these artifacts in order to increase accessibility and support important tasks such as impact analysis, system comprehension, system validation, and software maintenance. However, current traceability approaches are generally infeasible or laborious to use in an industry setting.;The software traceability state--of--the--art is advanced by this dissertation in the following ways. First, a novel framework for analyzing the difficulties of achieving software traceability is presented. This analysis framework reveals that traceability is difficult to achieve in practice because hindering factors from economic, technical, and social perspectives are intertwined. Second, important traceability insights are gleaned from both the development of a successful traceability tool in an industrial automation company and research areas such as e--Science, software architectures, and open hypermedia. Third, a novel automated technique for capturing software traceability links, called Architecture--Centric Traceability for Stakeholders (ACTS), is presented.;ACTS holistically attacks the multi--faceted traceability problem by centering all traceability links to the architecture, giving stakeholders greater control over the capture and usage of traceability links, and employing a combination of mechanisms (prospective capture, open hypermedia, and rules) that support the capture, management, and traversal of traceability links. Prospective link generation enables the capture of links in situ, while artifacts are generated or modified. This technique minimizes the overhead of capturing links and also nicely complements existing automated trace recovery techniques. Open hypermedia techniques enable links across heterogeneous and distributed artifacts, while rules enable the automatic capture of link semantics. In essence, ACTS is a technical framework that considers the economic and social challenges to traceability.;ACTS is validated in three ways. The utility of architecture--centric links is validated through a case study in software acquisition research. Support for stakeholder--driven traceability and for the capture of semantically--rich links across heterogeneous artifacts is validated through an exemplar implementation. Finally, improved accessibility to related artifacts is validated through experiments. |
