Trace Link Recovery for Architecture Decision Records (ADRs)
|
Typ | Bachelorarbeit oder Masterarbeit |
|---|---|---|
| Aushang | ||
| Betreuer | Wenden Sie sich bei Interesse oder Fragen bitte an: Jan Keim (E-Mail: jan.keim@kit.edu, Telefon: +49-721-608-45994) |
Architecture Decision Records (ADRs) have emerged as a lightweight and structured way to document architectural decisions throughout the lifecycle of a software system. Despite their growing adoption, ADRs are often disconnected from the actual system artifacts they influence—such as source code, models, or other documentation. This disconnect can limit the usefulness of ADRs in change impact analysis, maintenance, and architecture conformance checking.
Establishing trace links between ADRs and system artifacts is critical to bridge this gap. However, such trace links are rarely recorded explicitly, and no standard datasets currently exist for this task.
This thesis aims to explore Trace Link Recovery (TLR) methods for relating ADRs to relevant system artifacts. The focus is on both dataset creation and exploratory application of trace recovery techniques, including manual and automated approaches.
Key Goals:
1. Dataset Creation:
- Identify real-world software projects that make use of ADRs (e.g., open-source repositories using Markdown-based ADRs).
- Manually establish trace links between ADRs and at least one of the following:
- Source code (classes, modules, files)
- Architecture/design models (if available)
- Supplementary documentation (e.g., requirements, README files)
2. Exploratory TLR:
- Apply or adapt IR- and ML-based TLR techniques to the created dataset.
- Compare automatic recovery performance with the manually established ground truth.
- Analyze challenges and limitations specific to ADRs (e.g., abstract language, forward-looking decisions).