43-EFT.WP.Data.DatasetCards v1.0 | Chapter 13 Privacy, Ethics & Compliance

Home ／ Docs-Technical WhitePaper ／ 43-EFT.WP.Data.DatasetCards v1.0

Chapter 13 Privacy, Ethics & Compliance

I. Chapter Purpose & Scope

II. Terminology & Dependencies

• Terminology source: EFT.WP.Core.Terms v1.0; this chapter only adds privacy/compliance-related fields and constraints.
• Dependent volumes: Data contract/export: Core.DataSpec v1.0; metrology & dimensional checks: Core.Metrology v1.0; splits/distribution: Chapter 11; labels/ontology: Chapter 8; uncertainty & error budget: Chapter 10; citation/version posture: Citation & Cross-Reference Spec v0.1.

III. Fields & Structure (Normative)

privacy:

policy: "no-PII" # no-PII | limited-PII | special-category

lawful_basis: ["consent","research"] # lawful bases (examples)

data_minimization: true

data_categories: ["telemetry","imagery","text"]

special_category_flags: [] # list if health/biometric/minor, etc.

deidentification:

methods: ["hash-id","mask-location","binning-time"]

k_anonymity: 10

l_diversity: 2

dp_epsilon: null # record if differential privacy applies

reidentification_risk:

posture: "low" # low | medium | high

evidence: ["sampled-adversary-test","linkage-check"]

retention:

policy: "min-necessary"

delete_after_days: 365

data_subject_rights:

access_export: true

rectification: true

erasure: true

objection: true

contact: "privacy@org.example"

ethics:

intended_use: ["academic","benchmark"]

prohibited_use: ["surveillance","biometric_identification"]

harm_mitigation:

bias_scan: ["class","region"]

human_review: true

safety_precedence: true

fairness:

axes: ["class","region"]

gap_metric: "abs_diff"

threshold: 0.05

compliance:

regions: ["EU-GDPR","US-CCPA","CN-DSL"] # examples; align with reality

data_transfer:

mechanisms: ["standard-clauses"] # cross-border transfer mechanisms

access_control:

roles: ["owner","maintainer","reader"]

enforcement: ["signed-url","token","ip-allowlist"]

incident_response:

contact: "security@org.example"

sla_hours: 72

audits:

schedule: "annual"

artifacts: ["pii-scan.txt","dpiA.md"]

see:

- "EFT.WP.Core.DataSpec v1.0:EXPORT"

- "EFT.WP.Core.Metrology v1.0:check_dim"

(privacy and ethics are conditionally required: present when PII/sensitive data or ethical disclosures apply; export references appear in export_manifest.references[].)

IV. Data Classification & Minimization

• Minimization: Collect only what is necessary for research/engineering goals; enumerate in data_categories and justify in provenance.
• Special categories: If biometrics/health/minors are involved, declare in special_category_flags[] and add stronger controls (tighter access/retention).
• Consistency with labels/ontology: If labels carry sensitive semantics, enable safety-first and no-PII in labels.guidelines.precedence.

V. Lawful Bases & Consent Management

• Record lawful bases in lawful_basis[] (e.g., consent, legitimate_interest, research);
• If consent is used, record scope / withdrawal mechanism / timestamp, and restrict regions in distribution.regional_compliance[]; withdrawals must trace through lineage.

VI. De-identification & Re-identification Risk

• Methods: hash-id, geo/time blurring (e.g., binning-time), precision truncation, perturbation/synthetic data; if applying DP, record dp_epsilon and scope.
• Assessment: Provide sampled-attack/linkage evidence; align reidentification_risk.posture with audit artifacts.
• Impact: Report downstream effects (e.g., accuracy drop) in Chapter 12; include as a component in uncertainty.components[] when applicable.

VII. Access Control & Governance

• Roles & mechanisms: Fix in access_control.roles/enforcement; list access logs and token revocation policies as export artifacts.
• Least privilege: Default read-only mirrors; write access limited to owner/maintainer. For outsourced/crowdsourced labeling, isolate and minimize exposure.

VIII. Regional Compliance & Cross-Border Transfers

• Enumerate frameworks in compliance.regions[] and keep consistent with distribution.regional_compliance[];
• Provide templates or IDs for transfer mechanisms (e.g., standard contractual clauses) in the export manifest.

IX. Incident Response & Audits

• Incident response: Fix incident_response.contact and sla_hours (e.g., 72-hour notification window).
• Periodic audits: Record audits.schedule (e.g., annual) and audit artifacts (DPIA, PII scans, access logs) under export_manifest.artifacts[] with sha256.

X. Linkage with Splits/Distribution and Metrology/Uncertainty

• Splits/Distribution: For sensitive data, splits must not induce leakage; reflect regional limits in distribution.regional_compliance[] and mirror lists.
• Metrology/Uncertainty: Privacy processing may add bias/noise; include as systematic/random components in uncertainty.components[]; all numeric items must pass check_dim.

XI. Machine-Readable Fragment (Drop-in)

privacy:

policy: "limited-PII"

lawful_basis: ["consent"]

data_minimization: true

data_categories: ["audio","text"]

deidentification:

methods: ["hash-id","clip-duration","additive-noise"]

k_anonymity: 20

reidentification_risk: {posture:"low", evidence:["sampled-adversary-test"]}

retention: {policy:"min-necessary", delete_after_days:180}

data_subject_rights:

access_export: true

rectification: true

erasure: true

contact: "privacy@org.example"

ethics:

intended_use: ["academic","benchmark"]

prohibited_use: ["surveillance"]

harm_mitigation: {bias_scan:["class","region"], human_review:true, safety_precedence:true}

fairness: {axes:["class","region"], gap_metric:"abs_diff", threshold:0.05}

compliance:

regions: ["EU-GDPR"]

data_transfer: {mechanisms:["standard-clauses"]}

access_control:

roles: ["owner","maintainer","reader"]

enforcement: ["signed-url","token"]

incident_response: {contact:"security@org.example", sla_hours:72}

audits: {schedule:"annual", artifacts:["pii-scan.txt","dpia.md"]}

see:

- "EFT.WP.Core.DataSpec v1.0:EXPORT"

- "EFT.WP.Core.Metrology v1.0:check_dim"

(Reference anchors align with export_manifest.references[], carrying Volume+Version+Anchor.)

XII. Coupling with Export Manifest (Normative)

export_manifest:

artifacts:

- {path:"compliance/pii-scan.txt", sha256:"..."}

- {path:"compliance/dpia.md", sha256:"..."}

references:

- "EFT.WP.Core.DataSpec v1.0:EXPORT"

- "EFT.WP.Core.Metrology v1.0:check_dim"

(Artifacts must be verifiable; no shortcodes/aliases; version and anchor required.)

XIII. Chapter Compliance Checklist

• For PII/sensitive data, privacy and ethics are enabled with lawful bases, de-identification methods, and re-identification evidence recorded.
• distribution.regional_compliance[] aligns with compliance.regions[]; cross-border mechanisms recorded in the export manifest.
• Access control, incident response, audit plan, and artifacts are registered and verifiable; frozen splits do not create privacy leakage.
• Bias/noise from privacy processing is included in Chapter 10 uncertainty; all numbers/units pass check_dim.