{
  "passed": true,
  "workbook": "C:\\edocument.repo\\benchmarkingmodels\\benchextractions.chatgpt\\-REPORT-005\\ChatGPT_Gemma_Qwen_Academic_Statistical_Appendix.xlsx",
  "workbook_sha256": "767156cf4d4e28c4f97cdfba472d21a778ca7107a06949bb09621b5816d9583c",
  "common_audit_json": "C:\\edocument.repo\\benchmarkingmodels\\benchextractions.chatgpt\\-REPORT-005\\Four_System_Common_PDF_Audit.json",
  "common_audit_json_sha256": "1f073d111036f9ee2ddeaaa6b60542bf4202777c6686a83f04fbf3c316d7bfd0",
  "checks": {
    "FS_08_Common_Overall_matches_json": {
      "passed": true,
      "issues": []
    },
    "FS_09_Common_Category_matches_json": {
      "passed": true,
      "issues": []
    },
    "FS_10_Common_Segment_matches_json": {
      "passed": true,
      "issues": []
    },
    "FS_11_Common_Pairs_matches_json": {
      "passed": true,
      "issues": []
    },
    "FS_12_Common_Document_matches_json": {
      "passed": true,
      "issues": []
    },
    "FS_13_Common_Alignment_matches_json": {
      "passed": true,
      "issues": []
    },
    "FS_01_System_Summary_matches_source_workbooks": {
      "passed": true,
      "issues": []
    },
    "FS_02_Category_matches_source_workbooks": {
      "passed": true,
      "issues": []
    },
    "FS_03_Hybrid_BERT_Segments_matches_source_workbooks": {
      "passed": true,
      "issues": []
    },
    "FS_06_Statistics_matches_fixed_seed_calculation": {
      "passed": true,
      "issues": []
    },
    "pre_merge_sheets_preserved": {
      "passed": true,
      "issues": []
    },
    "source_files_exist": {
      "passed": true,
      "issues": []
    }
  },
  "source_hashes": [
    {
      "label": "Three-system comparative audit",
      "path": "C:\\edocument.repo\\benchmarkingmodels\\benchextractions.chatgpt\\-REPORT-005\\ChatGPT_Gemma_Qwen_24_Category_Comparative_Audit.xlsx",
      "exists": true,
      "sha256": "4c372c6f8d6b4b9dc4bc3d2cd7b4731ad17e0b08215641f70eb053083f55d34d"
    },
    {
      "label": "BERT combined workbook",
      "path": "C:\\edocument.repo\\benchmarkingmodels\\benchextractions.bert\\-REPORT-001\\BERT_CV_Extraction_24_Category_Combined_Report.xlsx",
      "exists": true,
      "sha256": "5c44a1d1cf56631129eecf8420e7c3205387294990c24f3fa3ff73b86fb74c49"
    },
    {
      "label": "BERT academic report",
      "path": "C:\\edocument.repo\\benchmarkingmodels\\benchextractions.bert\\-REPORT-001\\BERT_CV_Extraction_24_Category_Academic_Report.docx",
      "exists": true,
      "sha256": "7904cb0420d1ded2ef9af0c36bd46536236aed068d3cab5cd3e571a226460c65"
    },
    {
      "label": "HNLP combined workbook",
      "path": "C:\\edocument.repo\\benchmarkingmodels\\benchextractions.hnlp\\-REPORT-001\\HNLP_CV_Extraction_24_Category_Combined_Report.xlsx",
      "exists": true,
      "sha256": "273e03301fa746992360276902eb941a03837ae5eba7c5d710a3d223ad9dfa28"
    },
    {
      "label": "HNLP academic report",
      "path": "C:\\edocument.repo\\benchmarkingmodels\\benchextractions.hnlp\\-REPORT-001\\HNLP_CV_Extraction_24_Category_Academic_Report.docx",
      "exists": true,
      "sha256": "50b061f9a7a4c49b41b96c62dcac12b50899e1486000187b95bcc6d0c4f44cc9"
    },
    {
      "label": "Hybrid strategy",
      "path": "C:\\edocument.repo\\edoc-260\\modules\\seniorconsultant-core\\src\\main\\java\\org\\seniorconsultant\\core\\extract\\HNlpExtractionStrategy.java",
      "exists": true,
      "sha256": "2251d8be4b7a15c533b9d73c5dbfbd41c9c43f98cd5be7f091156059fd54fd58"
    },
    {
      "label": "Hybrid extractor",
      "path": "C:\\edocument.repo\\edoc-260\\modules\\seniorconsultant-core\\src\\main\\java\\org\\seniorconsultant\\core\\nlp\\HNlpExtractor.java",
      "exists": true,
      "sha256": "c95d8ce22cc3bd696c5914871de3bc1690f4a7daf3742eb82b8c04fcbf1b5fa0"
    },
    {
      "label": "BERT strategy",
      "path": "C:\\edocument.repo\\edoc-260\\modules\\seniorconsultant-core\\src\\main\\java\\org\\seniorconsultant\\core\\extract\\DjlTokenClassificationExtractionStrategy.java",
      "exists": true,
      "sha256": "bc883f625f74c5a227ac5951b24e17f63f390aad79221da963f3d0b2dfb2566f"
    },
    {
      "label": "BERT multilingual preset",
      "path": "C:\\edocument.repo\\edoc-260\\modules\\seniorconsultant-core\\src\\main\\java\\org\\seniorconsultant\\core\\extract\\DjlTokenClassificationMultilingualExtractionStrategy.java",
      "exists": true,
      "sha256": "c467394d018e1359ad83a2103e19e2a6fcb3a725fd2c34e24e3070dd3861416d"
    },
    {
      "label": "BERT runtime",
      "path": "C:\\edocument.repo\\edoc-260\\modules\\seniorconsultant-core\\src\\main\\java\\org\\seniorconsultant\\core\\extract\\DjlTokenClassificationRuntime.java",
      "exists": true,
      "sha256": "0e335bce8dc053638d801de9a5057ff9ea881c97dbbdc5c1583f32799822e272"
    },
    {
      "label": "BERT schema mapper",
      "path": "C:\\edocument.repo\\edoc-260\\modules\\seniorconsultant-core\\src\\main\\java\\org\\seniorconsultant\\core\\extract\\DjlSchemaMapper.java",
      "exists": true,
      "sha256": "4fdbe2bbec188fd8778a14d0b79f83fec1e86c7797541f73d13f6421a39dd752"
    },
    {
      "label": "BERT plan builder",
      "path": "C:\\edocument.repo\\edoc-260\\modules\\seniorconsultant-core\\src\\main\\java\\org\\seniorconsultant\\core\\extract\\SchemaDrivenEntityPlanBuilder.java",
      "exists": true,
      "sha256": "a932cf10f2a1501fb756f181f622efec3254c0b85fedf6f1e1b0e1c150e44318"
    }
  ],
  "provenance_rows": [
    {
      "Report/workbook material": "Figures 20\u201329; Tables 24C\u201324D; sheets FS_07_Common_Method\u2013FS_13_Common_Alignment",
      "Evidence class": "Empirical\u2014primary common audit",
      "Trace": "Four_System_Common_PDF_Audit.json; regenerated from 2,484 PDF files and four extractor ZIP sets",
      "Permitted interpretation": "Four-system comparison under the frozen lexical/n-gram PDF-support rule",
      "Restriction": "Not human semantic adjudication; thresholds and PDF text extraction remain measurement assumptions",
      "Audit status": "TRACEABLE"
    },
    {
      "Report/workbook material": "Figures 3\u20137, 9\u201310; Tables 5\u20138, 15; preserved base sheets",
      "Evidence class": "Empirical\u2014historical sensitivity layer",
      "Trace": "Pre-merge statistical appendix, preserved byte-for-value in the current workbook",
      "Permitted interpretation": "Historical Gemma/Qwen results under their documented earlier evaluator",
      "Restriction": "Must not be silently pooled with the common four-system evaluator",
      "Audit status": "TRACEABLE / SEPARATE TIER"
    },
    {
      "Report/workbook material": "Figure 19; Tables 24\u201326; sheets FS_01\u2013FS_06",
      "Evidence class": "Empirical\u2014source-workbook layer",
      "Trace": "Gemma/Qwen historical sheets plus BERT and HNLP combined workbooks; fixed-seed calculations in merge script",
      "Permitted interpretation": "Descriptive cross-tier profile; paired HNLP\u2013BERT inference only within their shared protocol",
      "Restriction": "Cross-tier values are not a universal five-system league table",
      "Audit status": "TRACEABLE / SEPARATE TIER"
    },
    {
      "Report/workbook material": "Figure 11; Table 13",
      "Evidence class": "Published architecture metadata",
      "Trace": "Official model cards/configuration plus supplied model filenames and launch commands",
      "Permitted interpretation": "Architecture description",
      "Restriction": "Parameter count is not a benchmark quality result",
      "Audit status": "SOURCE-VERIFIED"
    },
    {
      "Report/workbook material": "Figures 12\u201315",
      "Evidence class": "Conceptual or proposed experiment",
      "Trace": "Mathematical/causal exposition only",
      "Permitted interpretation": "Hypotheses and study-design requirements",
      "Restriction": "No router trace, path coefficient, reasoning-mode effect, or factorial outcome was measured",
      "Audit status": "NON-EMPIRICAL\u2014EXPLICITLY LABELLED"
    },
    {
      "Report/workbook material": "Figures 16\u201318; Tables 20\u201323; sheet FS_04_Architecture",
      "Evidence class": "Code-derived implementation analysis",
      "Trace": "Supplied edoc-260 Java source files with SHA-256 hashes",
      "Permitted interpretation": "Pipeline construction, prerequisites, and possible failure mechanisms",
      "Restriction": "No numeric sensitivity score or causal effect is inferred from code structure",
      "Audit status": "NON-EMPIRICAL\u2014EXPLICITLY LABELLED"
    },
    {
      "Report/workbook material": "Figure 17",
      "Evidence class": "Code-derived prerequisite diagram",
      "Trace": "HNLP and BERT Java code paths",
      "Permitted interpretation": "Conditions that must be controlled or recorded",
      "Restriction": "The former manually scored heatmap was removed; no Low/Moderate/High/Critical values remain",
      "Audit status": "UNSUPPORTED SCORES REMOVED"
    },
    {
      "Report/workbook material": "Figure 30; Tables 23A\u201323C",
      "Evidence class": "Method and validity framework",
      "Trace": "Implemented evaluator workflow and explicit validity argument",
      "Permitted interpretation": "How results were produced and what they can support",
      "Restriction": "Workflow boxes are not numerical outcomes",
      "Audit status": "REPRODUCIBLE METHOD"
    },
    {
      "Report/workbook material": "Table 31 and deployment conclusion",
      "Evidence class": "Mechanism-based deployment hypotheses",
      "Trace": "Architecture analysis",
      "Permitted interpretation": "Design hypotheses requiring controlled latency and external-validation studies",
      "Restriction": "No speed or cross-dataset stability ordering is claimed",
      "Audit status": "UNMEASURED RANKINGS REMOVED"
    }
  ],
  "removed_or_relabelled": [
    "Figure 17 manual ordinal sensitivity matrix",
    "Web stage-retention percentages that were illustrative rather than observed",
    "Web deployment coordinates that were manually positioned rather than measured",
    "Unmeasured speed ordering",
    "Unmeasured repeated-run ordering",
    "Unmeasured cross-dataset stability ordering"
  ]
}