Decision-Scoped Requirements Analysis: Synthetic Cases and Formative AI Question Traces

This dataset provides anonymized research artifacts supporting an exploratory study of AI-assisted iterative requirements analysis and decision-scoped analysis sufficiency. The package contains independently constructed fictional requirements scenarios, AI question-generation prompts, manually preserved response records, descriptive summaries, and validation scripts. Four exploratory ChatGPT runs produced 8, 8, 8, and 0 candidate clarification questions, respectively. The study investigates how AI-generated questions may help engineers identify unresolved assumptions, distinguish observed facts from unsupported conclusions, clarify responsibility boundaries, and examine the justification required for progression to conceptual logical-architecture review. The artifacts are intended for exploratory research and methodological transparency. The initial observations do not establish defect-detection accuracy, reliable sufficiency judgments, or improvements over human requirements analysis. Independent reference judgments and controlled human comparisons remain future work. All scenarios are fictional. No proprietary industrial requirements or product data are included. The package documents limitations concerning manual transcription, incomplete execution logs, and reproducibility.