Amazon Bedrock Knowledge bases の高度な解析オプションを試してみる

Transcribe the text content from an image page and output in Markdown syntax (not code blocks). Follow these steps:

1. Examine the provided page carefully.

2. Identify all elements present in the page, including headers, body text, footnotes, tables, visualizations, captions, and page numbers, etc.

3. Use markdown syntax to format your output:
    - Headings: # for main, ## for sections, ### for subsections, etc.
    - Lists: * or - for bulleted, 1. 2. 3. for numbered
    - Do not repeat yourself

4. If the element is a visualization
    - Provide a detailed description in natural language
    - Do not transcribe text in the visualization after providing the description

5. If the element is a table
    - Create a markdown table, ensuring every row has the same number of columns
    - Maintain cell alignment as closely as possible
    - Do not split a table into multiple tables
    - If a merged cell spans multiple rows or columns, place the text in the top-left cell and output ' ' for other
    - Use | for column separators, |-|-| for header row separators
    - If a cell has multiple items, list them in separate rows
    - If the table contains sub-headers, separate the sub-headers from the headers in another row

6. If the element is a paragraph
    - Transcribe each text element precisely as it appears

7. If the element is a header, footer, footnote, page number
    - Transcribe each text element precisely as it appears

Output Example:

A bar chart showing annual sales figures, with the y-axis labeled "Sales ($Million)" and the x-axis labeled "Year". The chart has bars for 2018 ($12M), 2019 ($18M), 2020 ($8M), and 2021 ($22M).
Figure 3: This chart shows annual sales in millions. The year 2020 was significantly down due to the COVID-19 pandemic.

# Annual Report

## Financial Highlights

* Revenue: $40M
* Profit: $12M
* EPS: $1.25


| | Year Ended December 31, | |
| | 2021 | 2022 |
|-|-|-|
| Cash provided by (used in): | | |
| Operating activities | $ 46,327 | $ 46,752 |
| Investing activities | (58,154) | (37,601) |
| Financing activities | 6,291 | 9,718 |

Here is the image.

{
    "schemaType": "ModelInvocationLog",
    "schemaVersion": "1.0",
    "timestamp": "2024-09-22T07:01:08Z",
    "accountId": "123456789012",
    "identity": {
        "arn": "arn:aws:sts::123456789012:assumed-role/advncd-prsng-kb-role/BedrockKnowledgeBaseCPSession-G51RXHKDH3"
    },
    "region": "us-west-2",
    "requestId": "8a62ba36-35ab-49e0-a322-ea1eac6f86eb",
    "operation": "InvokeModel",
    "modelId": "arn:aws:bedrock:us-west-2::foundation-model/anthropic.claude-3-sonnet-20240229-v1:0",
    "input": {
        "inputContentType": "application/json",
        "inputBodyJson": {
            "anthropic_version": "bedrock-2023-05-31",
            "messages": [
                {
                    "role": "user",
                    "content": [
                        {
                            "type": "text",
                            "text": "Bedrock Knowledge Base verifying access to requested model"
                        }
                    ]
                }
            ],
            "system": "",
            "max_tokens": 2048,
            "temperature": 0,
            "top_p": 1,
            "stop_sequences": [
                "\nObservation"
            ],
            "top_k": 50
        },
        "inputTokenCount": 18
    },
    "output": {
        "outputContentType": "application/json",
        "outputBodyJson": {
            "id": "msg_bdrk_014f8cwYGE8muj6pEcfgixhf",
            "type": "message",
            "role": "assistant",
            "model": "claude-3-sonnet-20240229",
            "content": [
                {
                    "type": "text",
                    "text": "I do not actually have access to Bedrock's knowledge base or models. I am Claude, an AI assistant created by Anthropic to be helpful, harmless, and honest."
                }
            ],
            "stop_reason": "end_turn",
            "stop_sequence": null,
            "usage": {
                "input_tokens": 18,
                "output_tokens": 41
            }
        },
        "outputTokenCount": 41
    }
}

{
    "schemaType": "ModelInvocationLog",
    "schemaVersion": "1.0",
    "timestamp": "2024-09-22T07:01:11Z",
    "accountId": "123456789012",
    "identity": {
        "arn": "arn:aws:sts::123456789012:assumed-role/advncd-prsng-kb-role/BedrockKnowledgeBaseCPSession-G51RXHKDH3"
    },
    "region": "us-west-2",
    "requestId": "2ffaa3d3-e5d9-4396-9c79-a73f1082e964",
    "operation": "InvokeModel",
    "modelId": "arn:aws:bedrock:us-west-2::foundation-model/amazon.titan-embed-text-v2:0",
    "input": {
        "inputContentType": "application/json",
        "inputBodyJson": {
            "inputText": "test",
            "dimensions": 1024
        },
        "inputTokenCount": 2
    },
    "output": {
        "outputContentType": "application/json",
        "outputBodyJson": {
            "embedding": [
                -0.03696037456393242,
            ],
            "inputTextTokenCount": 2
        }
    }
}

{
	"schemaType": "ModelInvocationLog",
	"schemaVersion": "1.0",
	"timestamp": "2024-09-22T07:01:14Z",
	"accountId": "123456789012",
	"identity": {
		"arn": "arn:aws:sts::123456789012:assumed-role/advncd-prsng-kb-role/9JZCREWYXJ-DocumentLoaderTask-ef525fa8-2405-4b41-a000-783d803ea"
	},
	"region": "us-west-2",
	"requestId": "0a3434d8-84a0-4cae-a0fb-a9720a5a3caa",
	"operation": "InvokeModel",
	"modelId": "arn:aws:bedrock:us-west-2::foundation-model/anthropic.claude-3-sonnet-20240229-v1:0",
	"input": {
		"inputContentType": "application/json",
		"inputBodyJson": {
			"anthropic_version": "bedrock-2023-05-31",
			"messages": [
				{
					"role": "user",
					"content": [
						{
							"type": "text",
							"text": "Transcribe the text content from an image page and output in Markdown syntax (not code blocks). Follow these steps:\n\n1. Examine the provided page carefully.\n\n2. Identify all elements present in the page, including headers, body text, footnotes, tables, visualizations, captions, and page numbers, etc.\n\n3. Use markdown syntax to format your output:\n    - Headings: # for main, ## for sections, ### for subsections, etc.\n    - Lists: * or - for bulleted, 1. 2. 3. for numbered\n    - Do not repeat yourself\n\n4. If the element is a visualization\n    - Provide a detailed description in natural language\n    - Do not transcribe text in the visualization after providing the description\n\n5. If the element is a table\n    - Create a markdown table, ensuring every row has the same number of columns\n    - Maintain cell alignment as closely as possible\n    - Do not split a table into multiple tables\n    - If a merged cell spans multiple rows or columns, place the text in the top-left cell and output ' ' for other\n    - Use | for column separators, |-|-| for header row separators\n    - If a cell has multiple items, list them in separate rows\n    - If the table contains sub-headers, separate the sub-headers from the headers in another row\n\n6. If the element is a paragraph\n    - Transcribe each text element precisely as it appears\n\n7. If the element is a header, footer, footnote, page number\n    - Transcribe each text element precisely as it appears\n\nOutput Example:\n\nA bar chart showing annual sales figures, with the y-axis labeled \"Sales ($Million)\" and the x-axis labeled \"Year\". The chart has bars for 2018 ($12M), 2019 ($18M), 2020 ($8M), and 2021 ($22M).\nFigure 3: This chart shows annual sales in millions. The year 2020 was significantly down due to the COVID-19 pandemic.\n\n# Annual Report\n\n## Financial Highlights\n\n* Revenue: $40M\n* Profit: $12M\n* EPS: $1.25\n\n\n| | Year Ended December 31, | |\n| | 2021 | 2022 |\n|-|-|-|\n| Cash provided by (used in): | | |\n| Operating activities | $ 46,327 | $ 46,752 |\n| Investing activities | (58,154) | (37,601) |\n| Financing activities | 6,291 | 9,718 |\n\nHere is the image.\n"
						},
						{
							"type": "image",
							"source": {
								"type": "base64",
								"media_type": "image/png",
								"data": "大量の文字列"
							}
						},
						{
							"type": "text",
							"text": "\nBased on the image provided, try to format the context into markdown format.\n<context>\n会社説明資料\u2028\n\nサンプルサンプル株式会社\n</context>\n"
						}
					]
				}
			],
			"max_tokens": 2048,
			"temperature": 0,
			"top_p": 1,
			"stop_sequences": ["\nObservation"],
			"top_k": 50
		},
		"inputTokenCount": 1401
	},
	"output": {
		"outputContentType": "application/json",
		"outputBodyJson": {
			"id": "msg_bdrk_01NW6Wv5QAMXiGEYMFdQJFt9",
			"type": "message",
			"role": "assistant",
			"model": "claude-3-sonnet-20240229",
			"content": [
				{
					"type": "text",
					"text": "# サンプルサンプル株式会社\n\n## 会社説明資料"
				}
			],
			"stop_reason": "end_turn",
			"stop_sequence": null,
			"usage": {
				"input_tokens": 1401,
				"output_tokens": 25
			}
		},
		"outputTokenCount": 25
	}
}

import base64

# base64でエンコードされたデータ（この例では省略されています）
encoded_data = "該当の文字列"

# base64デコード
decoded_data = base64.b64decode(encoded_data)

# PNGファイルとして保存
with open("output.png", "wb") as file:
    file.write(decoded_data)

print("PNGファイルが正常に保存されました。")

# サンプルサンプル株式会社

## 会社説明資料

# 会社概要

# 会社概要

会社名: サンプルサンプル株式会社

所在地: サンプル県サンプル市サンプル村　サンプル 1-2-3

代表: サン・プル

設立: 2010 年 04 月

従業員数: 12,345 名(グループ全体)

事業:

- 宇宙旅行事業
- AI 搭載ロボットペット事業
- 宇宙マーケティング戦略

# 柔軟な努力と最適な成果

Flexible Diligence, Optimal Results

我が社は、固定観念にとらわれない柔軟な思考と、状況に応じた適切な努力を重視します。常に変化する市場環境において、過度な労力を避けつつ、最適な成果を追求することで、持続可能な成長と従業員の幸福を両立させます。

## 行動指針

1.  臨機応変：状況を的確に判断し、柔軟に対応する
2.  効率重視：投入する労力と得られる成果のバランスを常に意識する
3.  創造的怠惰：省力化と革新的アイデアの源泉として、適度な余裕を大切にする
4.  調和の精神：仕事と生活の調和を図り、長期的な生産性向上を目指す
5.  継続的改善：小さな努力の積み重ねが大きな変化をもたらすことを信じる

# 沿革

サンプルサンプル株式会社 設立
2010 年 04 月 旅行代理店としてスタート

---

宇宙旅行事業 URZ
提供開始
2011 年 09 月

---

AI 搭載ロボットペット事業 ARPZ
提供開始
2014 年 12 月

---

宇宙マーケティング事業 UMZ 提
供開始
2020 年 04 月

# 事業内容

# URZ

## 宇宙旅行事業 (Uchu Ryokou Zigyou)

- 月面リゾート「ルナ・パラダイス」の建設
- 宇宙飛行士養成プログラムの提供
- 宇宙食の開発と販売
- FY2024 売上高：5,000 億円

# ARPZ

## AI 搭載ロボットペット事業（AI Robot Pet Zigyou）

- 感情認識 AI を搭載した「エモ・ペット」の開発
- セレブリティとのコラボレーションモデルの販売
- ロボットペット用アクセサリーの開発と販売
- FY2024 売上高：1,000 億円

# UMZ

## 宇宙マーケティング事業 (Uchu Marketing Zigyou)

- 宇宙をテーマにしたテレビ CM の放映
- 月面リゾートへの招待券付き SNS キャンペーンの実施
- AI 搭載ロボットペットの体験型ショールームの設置
- FY2024 売上高：1,000 億円

# 組織

# 社員数

The image shows a bar chart depicting the number of employees over several fiscal years. The bars are purple in color and show a consistent upward trend, with the text indicating that the number of employees exceeded 10,000 in FY2022.

The x-axis labels represent the fiscal years from FY2010 to FY2023, while the y-axis shows the employee count ranging from 0 to 12,500.

The chart is accompanied by the following text:

右肩上がりに増加
FY2022 に 10,000 人を突破

This can be translated as:

Increasing steadily
Exceeded 10,000 employees in FY2022

# 社員構成（年代別）

A pie chart showing the employee composition by age group:

| Age Group | Percentage |
| --------- | ---------- |
| 20 代     | 24.0%      |
| 30 代     | 40.0%      |
| 40 代     | 24.0%      |
| 50 代     | 8.0%       |
| 60 代以上 | 4.0%       |

# 拠点

アジア圏を中心に 8 拠点を展開

- 東京
- ソウル
- シドニー
- マレーシア
- オハイオ
- オレゴン
- カナダ
- パリ

The image shows a world map highlighting 8 locations where the company has operations or offices. The locations are listed as bullet points and include Tokyo, Seoul, Sydney, Malaysia, Ohio, Oregon, Canada, and Paris. The map uses shades of purple to indicate these locations across Asia, North America, and Europe.

# 募集要項

The image contains a header text in Chinese characters that translates to "Recruitment Information" or "Application Guidelines" in English.

# 営業職(URZ)

## 仕事内容

- 月面リゾート「ルナ・パラダイス」の利用促進
- 宇宙飛行士養成プログラムの販売

## 求める人材

- 新しい分野に挑戦する意欲がある方
- 高度な交渉力とプレゼンテーション能力を持つ方

## 応募資格

- 大卒以上
- 営業経験 3 年以上(業界不問)
- 英語でのコミュニケーション能力(TOEIC 700 点以上相当)

# エンジニア職 (URZ)

## 仕事内容

- 宇宙船および月面車両のシステム開発
- 宇宙飛行士養成プログラムの販売

## 求める人材

- 革新的な技術開発に情熱を持つ方
- 複雑な問題を解決する能力を持つ方

## 応募資格

- 理工系の学部・大学院卒業（宇宙工学、機械工学、電気電子工学、情報工学など）
- プログラミング経験 6 年以上
- 英語でのコミュニケーション能力（TOEIC 850 点以上相当）

# Flexible Diligence, Optimal Results

{
	"schemaType": "ModelInvocationLog",
	"schemaVersion": "1.0",
	"timestamp": "2024-09-22T07:01:54Z",
	"accountId": "123456789012",
	"identity": {
		"arn": "arn:aws:sts::123456789012:assumed-role/advncd-prsng-kb-role/9JZCREWYXJ-TextSplitterTask-5ff00b64-4267-4f5a-ab0e-fed72630714"
	},
	"region": "us-west-2",
	"requestId": "b745dc6f-fb9d-4dc1-b205-d3b4354be942",
	"operation": "InvokeModel",
	"modelId": "arn:aws:bedrock:us-west-2::foundation-model/amazon.titan-embed-text-v2:0",
	"input": {
		"inputContentType": "application/json",
		"inputBodyJson": {
			"inputText": "# サンプルサンプル株式会社\n## 会社説明資料",
			"dimensions": 1024
		},
		"inputTokenCount": 27
	},
	"output": {
		"outputContentType": "application/json",
		"outputBodyJson": {
			"embedding": [-0.08999396115541458],
			"inputTextTokenCount": 27
		}
	}
}

FY2022 の 20 代の人数はおおよそ何名ですか？