{
  "schema": "silktrain-pilot-sample/v1",
  "synthetic": true,
  "locale": "zh-hant",
  "workflow": "document-text",
  "service": "data-documents",
  "title": "文件與文字標註",
  "source_scenario": "掃描報告內有版次欄、表格及跨越兩行的命名實體。",
  "expected_delivery": "OCR 校訂、實體範圍、欄位值與原頁座標，依約定格式匯出。",
  "acceptance_check": "對照原頁核對欄位，檢查實體邊界與表格對齊。",
  "scoping_checklist": "準備具代表性的樣本、類別定義、邊界情況及標明單位或座標的匯出結構。",
  "examples": [
    {
      "id": "DOC-001",
      "source": "Weld W-07. Revision B.",
      "field": "weld_id",
      "start_char": 5,
      "end_char_exclusive": 9,
      "value": "W-07",
      "page": 1
    },
    {
      "id": "DOC-002",
      "source": "Weld W-07. Revision B.",
      "field": "revision",
      "start_char": 20,
      "end_char_exclusive": 21,
      "value": "B",
      "page": 1
    },
    {
      "id": "DOC-003",
      "source": "Review status: pending.",
      "field": "status",
      "start_char": 15,
      "end_char_exclusive": 22,
      "value": "pending",
      "page": 2
    }
  ]
}