52-Dataset Card Template v1.0 | Chapter 0 — Preface & User Guide

Home ／ Docs-Technical WhitePaper ／ 52-Dataset Card Template v1.0

Chapter 0 — Preface & User Guide

I. Purpose & Scope

• Standardize acquisition, structure, quality, and release of datasets, covering provenance/lineage, schema & units/dimensions, splits & versioning/freshness, quality gates & integrity, uncertainty & covariance, bias/ethics/privacy, interfaces & access, and benchmarking & comparative scoring.
• Applicable to batch, streaming, and near-real-time datasets; for outputs involving path quantities (arrival time/phase), the text must explicitly show gamma(ell) and d ell, and the data side records delta_form ∈ {general, factored}; publication requires p_dim = 1.0.

II. Audience & Use

• Audience: acquisition/experiment leads, data engineering & governance, metrology & review, benchmarking & release owners, interface/system integration.
• Use: follow chapter-by-chapter field/process specs; produce the DS_EXPORT/ directory and machine-readable manifests for cross-volume reference by Pipeline/Parameter/Error/Protocol cards.

III. Normative References

• See "EFT.WP.Core.Terms v1.0" P10-*, See "EFT.WP.Core.Equations v1.1" Ch.2 S20-1/S21-2, See "EFT.WP.Core.Metrology v1.0" check_dim, See "EFT.WP.Core.DataSpec v1.0" TARR.
• See "Parameter Registration Card Template v1.0" Ch.9, See "Error Budget Card Template v1.0" Ch.5/Ch.6/Ch.8/Ch.9, See "Pipeline Card Template v1.0" Ch.12/Ch.9/Ch.11, See "Experimental Protocol Card Template v1.0" Ch.5.
• Citations must use “volume + version + anchor (P/S/M/I)”, anchor coverage ≥ 90%, no external links/aliases; public releases cite v1.* only.

IV. Style & Numbering

• Section headings in bold Roman numerals; bullets with “- space”; ordered lists with “1. space”.
• Wrap inline symbols with backticks (e.g., T_arr, Phi, n_eff, c_ref); any expression with division/integrals/composites must use parentheses and declare path & measure.
• Conflict names: T_fil (tension) ≠ T_trans (transmittance); n (number density) ≠ n_eff (effective refractive index); forbid bare c, T, n.
• Unified arrival-time forms:
  T_arr = ( 1 / c_ref ) * ( ∫ n_eff d ell )
  T_arr = ( ∫ ( n_eff / c_ref ) d ell )

V. Deliverables & Layout

1. Required:
   • Figures: scale & distribution, missingness & quality, path profiles & intervals, benchmarks & comparative scoring.
   • Tables: field schema snapshot, G1–G8 compliance, key metrics & intervals, splits & versions.
   • Reports & manifests: schema.json, contract.yaml, split.yaml, validate_report.json, check_dim_report.json, privacy_policy.yaml (if applicable), audit.jsonl, report_manifest.yaml, signature SIGNATURE.asc.
2. Recommended layout: DS_EXPORT/figs|tables|reports|manifests/; all numeric outputs show units, and any division/integral/composite expression must use parentheses.

VI. Gates & Stops

• G1 Schema completeness | G2 Citation compliance | G3 Path conventions | G4 Dimensional closure | G5 Freshness | G6 Coverage conventions | G7 Covariance consistency | G8 Uniqueness.
• Triggering S1–S5 (dimensional failure/freshness failure/path missing/covariance mismatch/citation non-compliance) blocks release; when necessary enter [Restricted] and publish qualitative trends & diagnostics only.

VII. Validation & Release Flow

• Run /validate to obtain G1–G8 status and stops_triggered; pass I70-dim_check (p_dim = 1.0).
• Generate report_manifest.yaml with checksums/signature; audit.jsonl records provenance, versions, splits, and all changes.
• Cross-check see[]/references[]/version and cov_group/coverage consistency with Parameter/Error/Pipeline/Protocol volumes.
• External release: cite v1.* only; anchor coverage ≥ 90%; tag [Restricted] for non-compliant outputs.