Source of Authority Layer v4: Reproducibility Package for "Post-Retrieval Governance for Conflict-Aware Retrieval-Augmented Generation"
Abstract
This reproducibility package supports the working paper Source of Authority: Post-Retrieval Governance for Conflict-Aware Retrieval-Augmented Generation — A Controlled Evaluation on Enterprise Commercial-Lighting Documentation (Nelson, J., 2026). It contains the empirical artifacts underlying the v4 evaluation reported in §5 of the paper. The package includes: REGISTRATION_v4.md and REGISTRATION_v4.sha256 — the hash-chained pre-registration document and its three-link SHA-256 hash chain (baseline → Correction 1: persona text → Correction 2: cost-cap raise), locked before any LLM call against the v4 corpus. CORPUS_TAGGED_FINAL.jsonl — the tagged corpus with Authority Contracts applied to each chunk. INGEST_MANIFEST.json — 100% file accounting (812 files manifested, 811 indexed). CORPUS_HEALTH.md — conflict-link analysis (9.68% chunk-level conflict rate). rig_a_answers.jsonl, rig_b_answers.jsonl, rig_d_prime_answers.jsonl — per-condition answer logs with retrieval traces and selector debug output across the 150-question evaluation. judge_input.md and judge_key.json — the blind judge packet delivered to the external grader, and the rig-identity key released to the grader only after grading was returned. lib/ — the v4 framework code (tagger, conflict analyzer, retrieval, intent classifier, selector, harness). audit/ — the six-phase audit deliverables verifying the numerical claims in §5. grading_outputs_v4/ — the grader's deliverables (per-answer CSV, summary CSV, B-vs-D′ deltas…
Read the paper
Related studies
- Knowledge Management Perspective of Generative Artificial Intelligence
- Navigating Generative Artificial Intelligence Promises and Perils for Knowledge and Creative Work
- Characterizing the evolving research on enterprise content management
- Large language models and multimodal foundation models for precision oncology
- From Generative AI to Extended Reality: Multidisciplinary Perspectives on the Challenges, Opportunities, and Future of Educational Computing
- Generative Artificial Intelligence: A New Engine for Advancing Environmental Science and Engineering
Go deeper in Evidence Hub
Evidence Hub adds a study data sheet (design, sample size, evidence level, key numbers), an AI summary and a chat grounded in the paper's text.