{
  "schemaVersion": "1.0",
  "updates": [],
  "companyNews": [],
  "id": "b918",
  "language": "ja",
  "name": "Unstructured",
  "originalName": "Unstructured",
  "library": "tools",
  "category": "モデル開発",
  "categoryKey": "模型开发",
  "categoryKeys": [
    "模型开发"
  ],
  "categories": [
    "モデル開発"
  ],
  "entryPoints": [],
  "summary": "オープンソース Python 部品で PDF・画像・業務文書を構造化要素へ分割し、LLM 検索や知識基盤の入力を準備します。",
  "introduction": "本項目は Unstructured のオープンソース前処理ライブラリです。形式別パーサー・レイアウト認識・OCR で文書を構造化要素へ変換します。開発環境と形式に応じた依存関係が必要です。ホスト型 Transform とは能力が異なり、最新ホストモデルの成果をそのまま適用できません。既定の軽量テレメトリーは公式手順で停止できます。",
  "capabilities": [],
  "company": {
    "details": {
      "facts": [
        {
          "key": "website",
          "label": "会社公式サイト",
          "value": "https://unstructured.io/"
        },
        {
          "key": "focus",
          "label": "主な事業",
          "value": "AI 向け文書変換・前処理技術を提供し、オープンソースと商用ホスト製品を展開します。"
        }
      ],
      "otherProducts": []
    },
    "name": "Unstructured Technologies, Inc.",
    "originalName": "Unstructured Technologies, Inc.",
    "country": {
      "codes": [
        "US"
      ],
      "names": [
        "アメリカ合衆国"
      ],
      "display": "アメリカ合衆国",
      "basis": "headquarters",
      "basisLabel": "会社本社"
    }
  },
  "officialUrl": "https://github.com/Unstructured-IO/unstructured",
  "profileUpdatedOn": "2026-10-21",
  "sources": [
    {
      "kind": "product",
      "url": "https://github.com/Unstructured-IO/unstructured"
    },
    {
      "kind": "documentation",
      "url": "https://github.com/Unstructured-IO/unstructured/blob/main/README.md"
    }
  ],
  "news": [],
  "links": {
    "interactive": "products/unstructured--b918.html",
    "reading": "read/ja/products/unstructured--b918.html",
    "markdown": "read/ja/products/unstructured--b918.md",
    "json": "read/ja/products/unstructured--b918.json"
  }
}