Assessing Computable Gaps in AI Governance for Children: Evidence-Mechanism-Governance-Indicator Modelling of UNICEF's Guidance on Artificial Intelligence and Children 3.0 Using the Graph-GAP Framework Dataset

This dataset releases a computable gap assessment data asset for child-centred AI governance, employing the Graph-GAP framework to structurally model UNICEF's Guidelines on Artificial Intelligence and Children 3.0. The dataset translates normative texts into traceable Evidence–Mechanism–Governance–Indicator (EMGI) chains: extracting governance assertions and risk dimensions from sentence-level/paragraph-level evidence fragments, establishing mappings between mechanism hypotheses and governance tools, and further forming auditable indicator definitions, scoring rules, and evidence anchors. The dataset's core contributions are: Translating abstract governance principles into operationalisable metric systems and gap calculation workflows; Representing evidence-governance dependency relationships via graph structures to support reproducibility, auditing, and extensibility; Providing standardised data formats (CSV/JSONL/SQLite/Graph formats) for research, regulatory sandboxes, and corporate compliance self-assessment. This dataset is applicable for: AI governance evaluation, Child Rights Impact Assessment, automated compliance auditing, GraphRAG/Retrieval-Augmented Reasoning (RAG), and computational policy text research.