Skip to content

References: Compliance, Explainability, and Audit

  1. Explainable Artificial Intelligence - Wikipedia - Defines XAI methods including LIME, SHAP, attention visualization, and post-hoc explainability — directly foundational for this chapter's treatment of explainability-by-design vs. post-hoc explainability, model cards, and the EU AI Act's explainability requirements for high-risk AI systems.

  2. Audit Trail - Wikipedia - Covers audit trail design including immutability requirements, tamper evidence, and retention policies — directly supporting this chapter's audit trail design section showing how context graph decision traces satisfy regulatory audit requirements as ordinary graph traversal queries.

  3. Regulatory Compliance - Wikipedia - Explains regulatory compliance frameworks across industries including financial services, healthcare, and employment law — foundational for this chapter's treatment of automated decision regulation and the range of compliance requirements that context graphs address structurally.

  4. Artificial Intelligence: A Modern Approach (4th ed.) - Stuart Russell, Peter Norvig - Pearson - Chapter 27 covers AI safety, ethics, and accountability frameworks including algorithmic fairness, bias detection, and the governance requirements for deployed AI systems — directly supporting this chapter's bias audit, fairness audit, and algorithmic accountability sections.

  5. Semantic Web for the Working Ontologist (3rd ed.) - Dean Allemang, James Hendler, Fabien Gandon - ACM Books - Chapter 11 covers data governance, provenance, and rights management using semantic web technologies — relevant to this chapter's GDPR right to explanation and data retention/purge policy sections where linked data provenance enables individual-level decision trace retrieval.

  6. General Data Protection Regulation - Wikipedia - Covers GDPR's Article 22 right to explanation for automated decisions, data subject rights, and data retention obligations — directly foundational for this chapter's GDPR explainability requirement and right to explanation sections.

  7. EU AI Act - European Commission - The official EU AI Act legislative text defining risk categories, requirements for high-risk AI systems, and logging/explanation obligations — directly relevant to this chapter's EU AI Act section showing how context graph decision traces satisfy the Act's technical documentation and logging requirements.

  8. Differential Privacy - Wikipedia - Explains differential privacy as a formal mechanism for protecting individual records in aggregate queries — supporting this chapter's fairness audit and bias detection sections where privacy-preserving analytics over decision trace populations must avoid revealing individual decisions.

  9. AI Alignment - Wikipedia - Covers AI alignment techniques including red teaming, safety testing, and robustness evaluation — directly supporting this chapter's AI red teaming section for testing context-graph-powered systems against adversarial inputs and edge cases.

  10. Information Security - Wikipedia - Covers security frameworks for protecting sensitive data and audit records — supporting this chapter's compliance gap analysis section and the data retention/purge policy design that must balance regulatory retention requirements against privacy obligations for decision trace data.