Verwendung von GraphRAG zur Erstellung von Ideen aus PDF-Dokumenten

• •Möglicherweise sind Ziel-API-Anmeldedaten erforderlich

{
  "id": "5wnJMLkwYeOXdFkE",
  "meta": {
    "instanceId": "2a26454b0172ffcb8d70ba77c235b1209f92cd71bf06c79ba609c7173b416d68",
    "templateCredsSetupCompleted": true
  },
  "name": "Generate Content Ideas from PDF Documents using GraphRAG",
  "tags": [
    {
      "id": "66wgFoDi9Xjl74M3",
      "name": "Support",
      "createdAt": "2025-05-21T17:06:32.355Z",
      "updatedAt": "2025-05-21T17:06:32.355Z"
    },
    {
      "id": "kRM0hQV2zw7VxrON",
      "name": "Research",
      "createdAt": "2025-05-21T19:44:19.136Z",
      "updatedAt": "2025-05-21T19:44:19.136Z"
    },
    {
      "id": "sJk9cUvmMU8FkJXv",
      "name": "AI",
      "createdAt": "2025-05-20T13:16:15.636Z",
      "updatedAt": "2025-05-20T13:16:15.636Z"
    }
  ],
  "nodes": [
    {
      "id": "212126f1-2a06-4211-814d-6b5a46df31ad",
      "name": "Datei in PDF konvertieren",
      "type": "n8n-nodes-base.httpRequest",
      "disabled": true,
      "position": [
        1880,
        180
      ],
      "parameters": {
        "url": "https://v2.convertapi.com/convert/pdf/to/txt",
        "method": "POST",
        "options": {
          "response": {
            "response": {
              "responseFormat": "text"
            }
          }
        },
        "sendBody": true,
        "contentType": "multipart-form-data",
        "sendHeaders": true,
        "authentication": "genericCredentialType",
        "bodyParameters": {
          "parameters": [
            {
              "name": "file",
              "parameterType": "formBinaryData",
              "inputDataFieldName": "data"
            }
          ]
        },
        "genericAuthType": "httpBearerAuth",
        "headerParameters": {
          "parameters": [
            {
              "name": "Accept",
              "value": "application/octet-stream"
            }
          ]
        }
      },
      "credentials": {
        "httpBearerAuth": {
          "id": "9fXf9Np7XsCWgxhg",
          "name": "Perplexity"
        }
      },
      "notesInFlow": true,
      "typeVersion": 4.2
    },
    {
      "id": "d0a606d8-c62e-42aa-8d82-e9adcc33f5f5",
      "name": "Haftnotiz5",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        1840,
        -400
      ],
      "parameters": {
        "color": 2,
        "width": 360,
        "height": 820,
        "content": "## Optional: Better PDF Conversion\n\n### Standard Map PDF to Text node will split your PDF files into very short chunks, which deteriorates retrieval. \n\nUse can use [ConvertAPI](https://convertapi.com?ref=4l54n) which is a high-quality convertor that will respect the layout of the original document and not cut the paragraphs into short chunks. \n\nHere is an HTTP node that makes a request to their API to convert the PDF into text. If you have a ConvertAPI account, you can replace the \"Extract Text from PDF\" node in Step 3 with this node. \n\nNote that you will need to map the text output from this node correctly in the Step 4 after.\n"
      },
      "typeVersion": 1
    },
    {
      "id": "f31d316f-d10a-47a4-b715-635cd58b996d",
      "name": "Bei Formularübermittlung",
      "type": "n8n-nodes-base.formTrigger",
      "position": [
        -380,
        -60
      ],
      "webhookId": "baea8339-e0c0-43e1-acdf-d41bb25b2b85",
      "parameters": {
        "options": {
          "appendAttribution": false
        },
        "formTitle": "Find Content Gaps in Your PDF Files",
        "formFields": {
          "values": [
            {
              "fieldType": "file",
              "fieldLabel": "Add Your Files",
              "acceptFileTypes": ".pdf"
            }
          ]
        },
        "formDescription": "Upload the files you'd like to analyze and we will extract content gaps and interesting questions based on them."
      },
      "typeVersion": 2.2
    },
    {
      "id": "56bae190-80e7-4fff-a06c-ee9ac74e5262",
      "name": "InfraNodus KI-Fragen",
      "type": "n8n-nodes-base.httpRequest",
      "disabled": true,
      "position": [
        960,
        200
      ],
      "parameters": {
        "url": "https://infranodus.com/api/v1/graphAndAdvice?doNotSave=true&optimize=develop&includeGraph=false&includeGraphSummary=true&gapDepth={{ $json.randomNum }}",
        "method": "POST",
        "options": {},
        "sendBody": true,
        "authentication": "genericCredentialType",
        "bodyParameters": {
          "parameters": [
            {
              "name": "aiTopics",
              "value": "true"
            },
            {
              "name": "requestMode",
              "value": "question"
            },
            {
              "name": "text",
              "value": "={{ $json.text }}"
            }
          ]
        },
        "genericAuthType": "httpBearerAuth"
      },
      "credentials": {
        "httpBearerAuth": {
          "id": "FPDx6PA5CtzGEIQc",
          "name": "InfraNodus DeeMeeTree API Key"
        }
      },
      "typeVersion": 4.2
    },
    {
      "id": "d546371b-c109-4273-9fb8-8abc5b89c4c1",
      "name": "Binärdateien in PDF konvertieren",
      "type": "n8n-nodes-base.code",
      "position": [
        -60,
        -60
      ],
      "parameters": {
        "jsCode": "let results = [];\n\nfor (let item of items) {\n    if (item.binary) {\n        // If there's binary data in the item, process each binary file\n        for (let key in item.binary) {\n            // Use the key as the file name\n            let binaryKey = key.replace(/\\s/g, '_'); // Replace spaces with underscores for the key\n            results.push({\n                json: {\n                    fileName: binaryKey\n                },\n                binary: {\n                    [binaryKey]: item.binary[key] // Use the modified key for the binary data\n                }\n            });\n        }\n    }\n}\n\nreturn results;\n"
      },
      "typeVersion": 2
    },
    {
      "id": "e4b0147b-c49e-4acc-90e8-976d6e0b177b",
      "name": "Text aus PDF-Dateien extrahieren",
      "type": "n8n-nodes-base.extractFromFile",
      "position": [
        280,
        -60
      ],
      "parameters": {
        "options": {},
        "operation": "pdf",
        "binaryPropertyName": "={{ $json.fileName }}"
      },
      "typeVersion": 1
    },
    {
      "id": "404f2c2b-0e97-4b49-96d4-16cb23c1d82e",
      "name": "Für InfraNodus vorbereiten",
      "type": "n8n-nodes-base.code",
      "position": [
        580,
        -60
      ],
      "parameters": {
        "jsCode": "\nlet plainText = '' // we send plain text from all the PDFs to InfraNodus for analysis\n\nconst randomNum = Math.floor(Math.random() * 3); // replace this with a 0 if you'd like to address the biggest gap in the knowledge graph\n\nfor (let item of items) {\n   plainText += item.json.text + '\\n\\n'  \n}\n\n\nreturn {text: plainText, randomNum};"
      },
      "typeVersion": 2
    },
    {
      "id": "8d12370c-aa4a-41ad-a3dc-39090779abd0",
      "name": "Im Formular dem Benutzer anzeigen",
      "type": "n8n-nodes-base.form",
      "position": [
        1380,
        -60
      ],
      "webhookId": "05cde493-a003-4979-9bf0-2b09c77a4044",
      "parameters": {
        "operation": "completion",
        "respondWith": "showText",
        "responseText": "=<br>\n<h3>{{ $json.aiAdvice[0].text }}</h3>\n<br>\n"
      },
      "typeVersion": 1
    },
    {
      "id": "d4b56485-ae4f-40a6-a69e-6362fa224cc5",
      "name": "InfraNodus GraphRAG KI-Empfehlung",
      "type": "n8n-nodes-base.httpRequest",
      "position": [
        960,
        0
      ],
      "parameters": {
        "url": "=https://infranodus.com/api/v1/graphAndAdvice?doNotSave=true&optimize=develop&includeGraph=false&includeGraphSummary=true&gapDepth={{ $json.randomNum }}",
        "method": "POST",
        "options": {},
        "sendBody": true,
        "authentication": "genericCredentialType",
        "bodyParameters": {
          "parameters": [
            {
              "name": "aiTopics",
              "value": "true"
            },
            {
              "name": "requestMode",
              "value": "idea"
            },
            {
              "name": "text",
              "value": "={{ $json.text }}"
            }
          ]
        },
        "genericAuthType": "httpBearerAuth"
      },
      "credentials": {
        "httpBearerAuth": {
          "id": "FPDx6PA5CtzGEIQc",
          "name": "InfraNodus DeeMeeTree API Key"
        }
      },
      "typeVersion": 4.2
    },
    {
      "id": "30da3a64-5020-44fb-8e56-38b68e66d342",
      "name": "Haftnotiz",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        -420,
        -400
      ],
      "parameters": {
        "height": 520,
        "content": "## Step 1: User uploads the PDF files for analysis\n\n### You can expose this endpoint and make it publicly available via a URL to your organization."
      },
      "typeVersion": 1
    },
    {
      "id": "90c0fb6f-cc11-43c7-bc06-2794968204ee",
      "name": "Haftnotiz1",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        -120,
        -400
      ],
      "parameters": {
        "width": 280,
        "height": 520,
        "content": "## Step 2: Convert uploaded binaries into PDF files\n\n### We need to convert the binaries uploaded to the PDF files so we can extract text from them."
      },
      "typeVersion": 1
    },
    {
      "id": "96d48573-5cb9-4ebe-8c64-ba7e14113a2e",
      "name": "Haftnotiz2",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        220,
        -400
      ],
      "parameters": {
        "width": 220,
        "height": 520,
        "content": "## Step 3: Extract plain text from PDF files\n\n### For better quality text extraction, you can use the optional [ConvertAPI](https://convertapi.com?ref=4l54n) node to the right, which respects the files' original formatting."
      },
      "typeVersion": 1
    },
    {
      "id": "8f1fcffa-1738-4e84-9552-742bb563ab9d",
      "name": "Haftnotiz3",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        520,
        -400
      ],
      "parameters": {
        "width": 220,
        "height": 520,
        "content": "## Step 4: Combine extracted text into a text string\n\n### Prepare data for InfraNodus: combine all the extracted text into a text string and also tell InfraNodus the gap depth it should use when generating advice"
      },
      "typeVersion": 1
    },
    {
      "id": "945f7df6-2b1e-44e9-ba89-0acf52b4b26e",
      "name": "Haftnotiz4",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        820,
        -400
      ],
      "parameters": {
        "width": 380,
        "height": 820,
        "content": "## Step 5: Use InfraNodus GraphRAG to Find a Content Gap and Generate Advice\n\n### Optionally, you can use the node below to generate questions (AI prompts) instead of ready-made ideas. These may be more interesting if you'd like to integrate this template into your existing research workflow.\n\n🚨 PROVIDE YOUR INFRANODUS API KEY HERE"
      },
      "typeVersion": 1
    },
    {
      "id": "3ffd5445-7e0b-43b5-85ab-790407ece21b",
      "name": "Haftnotiz6",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        1280,
        -400
      ],
      "parameters": {
        "width": 380,
        "height": 820,
        "content": "## Step 6: Show advice / question / prompt to the user\n\n### Optionally, you can feed the response to your other n8n workflow or expose it via a webhook and show it in your own app using an iframe."
      },
      "typeVersion": 1
    },
    {
      "id": "9320b872-4eea-4b70-942f-ec2688d250de",
      "name": "Haftnotiz7",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        -420,
        180
      ],
      "parameters": {
        "color": 5,
        "width": 1160,
        "height": 1000,
        "content": "# Find content gaps in PDF documents using the InfraNodus GraphRAG\n\n## How does the [InfraNodus](https://infranodus.com) GraphRAG work?\n\n### 1. It represents your text as a network of concepts and relations\n\n### 2. It then identifies the clusters that are furthest apart from each other — they appear in the same context (your texts) but are not well connected\n\n### 3. InfraNodus will then use the AI to generate an idea / question / prompt that bridges this gap — touching upon relevant topics but connecting them in a new way\n\n![structural gap infranodus](https://infranodus.com/images/front/infranodus-structural-gaps-ideas.jpg)"
      },
      "typeVersion": 1
    }
  ],
  "active": false,
  "pinData": {},
  "settings": {
    "executionOrder": "v1"
  },
  "versionId": "4cacd2c9-73cf-433d-ab20-a1226355c502",
  "connections": {
    "f31d316f-d10a-47a4-b715-635cd58b996d": {
      "main": [
        [
          {
            "node": "d546371b-c109-4273-9fb8-8abc5b89c4c1",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "404f2c2b-0e97-4b49-96d4-16cb23c1d82e": {
      "main": [
        [
          {
            "node": "d4b56485-ae4f-40a6-a69e-6362fa224cc5",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "d546371b-c109-4273-9fb8-8abc5b89c4c1": {
      "main": [
        [
          {
            "node": "e4b0147b-c49e-4acc-90e8-976d6e0b177b",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "e4b0147b-c49e-4acc-90e8-976d6e0b177b": {
      "main": [
        [
          {
            "node": "404f2c2b-0e97-4b49-96d4-16cb23c1d82e",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "d4b56485-ae4f-40a6-a69e-6362fa224cc5": {
      "main": [
        [
          {
            "node": "8d12370c-aa4a-41ad-a3dc-39090779abd0",
            "type": "main",
            "index": 0
          }
        ]
      ]
    }
  }
}