Researchers have developed and evaluated seven configurations for ranking non-functional requirement (NFR) coverage gaps in software development. These configurations utilize statistical, embedding-based, and large language model (LLM) approaches, drawing evidence from requirement statements, peer requirements, and category statistics. Experiments on the PROMISE dataset demonstrated that configurations using peer category statistics performed similarly well, with direct LLM ranking showing the largest evidence gain. The study found that structured peer evidence is more crucial than ranker sophistication for count-changing category-level gaps, suggesting these rankings should guide analyst inspection rather than automate completeness judgments. AI
IMPACT This research offers improved methods for identifying missing non-functional requirements in software projects, potentially enhancing software quality and development efficiency.
RANK_REASON The cluster contains an academic paper detailing a new methodology for software requirements analysis. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →