[2/3] Quality Metrics in Software Architecture (Category Architecture)
Continue. story About this scientific article, we will talk about the methodology of research and the approach to implementation.
To implement their framework, the authors took the following tools The model is implemented using a domain-specific language (Domain-Specific Language, DSL) xText-based, which allows you to textually describe elements of quality and their relationship. A code generator is used for visualization. (xTend)It builds interactive web interfaces: lists of metrics and attributes, and a graph of the relationships between them. The directory supports expansion and teamwork: any participant can propose a new metric or attribute through pull request, after which the data automatically enters the visualization. (There were no pull requests in the repository all the time.)
If you describe the approach of the authors to the study, it was quite strict.
- They formulated research questions. (RQ1–RQ4)What metrics and attributes are used, whether they are universal or specific, internal or external, and how to apply them in practice. They formulated a search strategy: keyword queries in the five largest scientific databases. (IEEExplore, ACM Digital Library, Scopus, Springer Link, ScienceDirect) period 2011–2022 Yeah.
- They filtered found publications by relevance, language, type of publication and other criteria, from 746 There are still publications 33 They analysed and classified the metrics and attributes found using ISO/IEC characteristics. 25010:2011And then they added generic/specific classification. (universal/specific) internal/external (internal) Based on this, they assembled a model and then used DSL to describe and generate visualizations.
This approach ensures completeness of coverage, transparency of selection criteria and reproducibility of results. An important aspect is the openness of the catalogue for further replenishment and collective development.
The authors emphasize that their catalog and framework is a “living” tool that should be developed together with the professional community. Key areas for future research and development: Evaluation and validation involving stakeholders: It is planned to conduct user research with architects and experts to test the usability and usefulness of the DSL and catalog. Community-drive development: the catalogue is open to external contributions, its development depends on the activity of the community of engineers and researchers. Integration into practice: the catalog can be used to: Identify new or missing relationships between metrics and attributes Support tools for automatic assessment of architecture quality; Analysis of the impact of some quality attributes on others (impact traceability); Building reusable quality models and SaaS solutions that do not depend on a specific architectural language. Expansion of the set of external metrics: most of the metrics found are internal, so more external metrics reflecting the perception of quality by users and customers are required. Validation of Attribute Relationships: The catalogue helps to identify unexpected or controversial relationships between attributes of quality, which may lead to new research and experimental verification.
As a result, this article offers a systematic and practically applicable catalog of metrics for assessing software architecture, implemented in the form of an extensible model and framework.
#Architecture #Metrics #Software #Engineering #SystemDesign #Management