Abstract
Decision-making performs an integral role in software architecture (SA). SA is considered the earliest design decisions regarding a software system and, therefore, bears enormous significance in its development and maintenance. Significant research has focused on the improvement of documentation practices and architectural knowledge (AK) to manage design decisions. However,However, research suggests that the potential of provenance has not been utilised in software architecture. Provenance is described as the record of the lineage of processes, information, artifacts and systems. Provenance has been adopted in several domains, including healthcare, emergency response research and education. Furthermore, the metadata can be used to track changes in processes and systems, therefore, improving traceability, accuracy and trust. This thesis proposes the SA_PROV model, which is a provenance-based model to capture architectural decision-making (ADM) processes. The research employs the design science research methodology (DSRM), encompassing the various stages of creating an artifact. The approach begins with an investigation into the existing literatureon architectural decision-making, provenance and other relevant areas to discover knowledge gaps, leading to the suggestion generation of ideas. The proposed idea, the SA_PROV model is developed and evaluated over a comparison with pioneering studies,
two ‘generate/test’ iterations with domain experts and a follow-up study. An analysis of ADM and AKM issues suggests that provenance data can improve accuracy, establish trust and provide opportunities for decision reuse. Also, the model can provide
opportunities for further advancements in developing tools for the visualisation of SADM.
| Date of Award | 16 Jun 2025 |
|---|---|
| Original language | English |
| Supervisor | Gary Allen (Main Supervisor) |