Documenting and Attributing Software Architecture Rationale in Software Architecture Models

Abstract

Despite many papers emphasizing the importance of recording software architecture design rationale documentation, there exists a research gap in the field in terms of systematic literature reviews about different models used to document design rationale. To address this gap, this thesis has the scope of a systematic literature review, taking in 21 source papers between years 2007-2024. We also wanted to find out about the possible benefits and drawbacks of different models, and the domains where they were used to get at least a surface level understanding of possible relationships between these things. The results were that ontologies and metamodels were the most popular models used, the most quoted benefits were traceability gains and increased understanding of architecture decisions, while most quoted drawbacks was the additional overhead required by these models. The different domains in which the models were tested did not yield any impactful findings, as the largest field was software with 3 quotations. Taking a look at the source papers, it also seems like the pace of papers being written about these topics has slowed down significantly over the years, with a slight resurgence again in 2024. Altogether, it seems that the main barrier for more widespread use of models in design rationale documentation is the overhead that they require in realms of documentation and learning to use them. Because of this, a possible future research direction could be reducing the overhead by either simplifying the models, or reducing the documentation effort needed to use the models efficiently.