Requirements Engineering
Reviewed: 2025-10-10
SERL advances requirements engineering research to facilitate requirements engineering endeavours of high quality. Requirements manifest themselves in various forms but they are mostly written in natural language. A core thread therefore defines what “good” requirements look like in practice, introducing an extensible ontology of quality factors, fitness-for-purpose activities, and ways to reason about quality evidence. Results also temper “smell hunting” with perceptions of real harm and highlight how process choices interact with stakeholder well-being.
Work on causality and conditionals shows how phrasing impacts ambiguity, downstream tests, and automation. Tooling (such as CiRA, CEREC) detects cause–effect relations and turns well-formed conditionals into acceptance-test descriptions, closing the gap from text to checks.
Traceability is reframed as knowledge organization. Domain taxonomies and recommenders (TT-RecS, taxonomic trace links) create early, maintainable traces and set quality attributes for robust taxonomies. This taxonomic lens extends to regulatory work-artefact-based approaches, sector-specific practices, and accessibility obligations make compliance more systematic and tool-supported.
Further, the research connects requirements to validation assets and product decisions: performance-requirements verification with test-environment generation, LLM-based labeling of recorded GUI tests, stakeholder-satisfaction–aware change prioritization, and scoping of quality requirements in product strategy. Complementary threads include data-centric and safety-critical contexts, global collaboration, personas, and feedback in continuous delivery.
Finally, one central thread for investigations position requirements as a central bridge between complex regulations and software engineering artefacts, thus, enabling regulatory compliance. We approach this in a more foundational, model-based fashion and in a tool-supported fashion, enabling regulatory compliance by-design and a systematic compliance assurance.
Together, our work turns requirements from static text into living, test-connected, and compliance-aware assets that steer delivery with evidence.
Current and Future Work
We expect lean, fitness-for-purpose dashboards powered by quality ontologies and Bayesian inference. NLP pipelines will operationalize causality/conditionals for ambiguity checks and automatic test generation. Taxonomy-driven traceability will anchor regulatory compliance across domains, from FinTech to accessibility. Stronger links between requirements and automated GUI tests will mature, alongside attribute sets and prioritization grounded in stakeholder value.