OPENAI_API_KEY

[abdullah91111]

2025-02-22 20:55:44 openai.AuthenticationError: Error code: 401 - {'error': {'message': 'Incorrect API key provided: openai_a**_key. You can find your API key at https://platform.openai.com/account/api-keys.', 'type': 'invalid_request_error', 'param': None, 'code': 'invalid_api_key'}}

When the document is processed the status shows failed and the above error occurs.

However I have updated the .env file with my open ai api key

If im correct I only need to replace OPENAI_API_KEY = "my_API_KEY" and nowhere else right?

If someone can let me know the why this error occurs.

[abdullah91111]

Also, since Difbott offers a free tier but is eventually a paid service, is the diffbott api key mandatory, and if so, for what?

[kaustubh-darekar]

Hi @abdullah91111
Openai key needs to be updated at at two places
OPENAI_API_KEY = #if you are using openai embedding model

For using LLM for extraction and chatbot, any of the following can be configured
LLM_MODEL_CONFIG_openai_gpt_3.5="gpt-3.5-turbo-0125,openai_api_key"
LLM_MODEL_CONFIG_openai_gpt_4o_mini="gpt-4o-mini-2024-07-18,openai_api_key"
LLM_MODEL_CONFIG_openai_gpt_4o="gpt-4o-2024-11-20,openai_api_key"
LLM_MODEL_CONFIG_openai-gpt-o3-mini="o3-mini-2025-01-31,openai_api_key"

For more clarity you can refer example.env in backend folder.

Diffbot is not mandatory if you are not using diffbot model.

[abdullah91111]

Another question: is it possible to use this as a rest API? If so is there any guide on how to do so

[kartikpersistent]

our api's are designed specific to our application so we will not recommend to use our Api's for budling the application

[kartikpersistent]

you can use neo4j graphrag python package for building the gen ai application

[abdullah91111]

In my use case scenario I dont want to build on top of the application but just use the rest api and integrate it with my n8n automated workflow. Would that be possible.

[kartikpersistent]

in that case you can use by providing appropriate payload

[abdullah91111]

Issue: Unable to Extract Entity Relationships from a Wikipedia Page

Context

I’m using the n8n HTTP node to integrate with GraphRAG. The process consists of two steps:

1. Uploading a Wikipedia document via /url/scan → Works Successfully ✅
2. Extracting entity relationships via /extract → Fails with a document node error ❌

---

Step 1: Uploading the Wikipedia Document via /url/scan

I send a POST request to /url/scan with these parameters:

Request Configuration

Method: POST

URL: http://backend:8000/url/scan

Content-Type: Form-Data

Body Parameters:

• uri: bolt://neo4j-apoc:7687
• userName: neo4j
• password: ***
• database: neo4j
• model: openai_gpt_4o
• wiki_query: https://en.wikipedia.org/wiki/Albert_Einstein
• source_type: Wikipedia

Response Output

{
  "status": "Success",
  "data": {
    "elapsed_api_time": "10.14"
  },
  "success_count": 1,
  "failed_count": 0,
  "message": "Source Node created successfully for source type: Wikipedia and source: https://en.wikipedia.org/wiki/Albert_Einstein",
  "file_name": [
    {
      "fileName": "Albert_Einstein",
      "fileSize": 8074,
      "url": "https://en.wikipedia.org/wiki/Albert_Einstein",
      "language": "en",
      "status": "Success"
    }
  ]
}


Step 2: Extracting Entity Relationships via /extract
After successfully uploading the Wikipedia document, I attempt to extract entity relationships using /extract.

Request Configuration
Method: POST
URL: http://backend:8000/extract
Content-Type: Form-Data
Body Parameters:
uri: bolt://neo4j-apoc:7687
userName: neo4j
password: ***
database: neo4j
model: openai_gpt_4o
file_name: Albert_Einstein
wiki_query: https://en.wikipedia.org/wiki/Albert_Einstein
source_type: Wikipedia
allowedNodes: Document, Chunk, Entity
allowedRelationship: PART_OF, HAS_ENTITY
token_chunk_size: 512
chunk_overlap: 52
language: en

### **Error Response**  
2025-02-27 19:17:00 2025-02-27 15:17:00,680 - Total Pages from Wikipedia = 24
2025-02-27 19:17:00 2025-02-27 15:17:00,712 - Time taken database connection: 0.03 seconds
2025-02-27 19:17:00 2025-02-27 15:17:00,783 - Index already exist,Skipping creation. Time taken: 0.07 seconds
2025-02-27 19:17:00 2025-02-27 15:17:00,783 - Break down file into chunks
2025-02-27 19:17:00 2025-02-27 15:17:00,784 - Split file into smaller chunks
2025-02-27 19:17:00 2025-02-27 15:17:00,812 - creating FIRST_CHUNK and NEXT_CHUNK relationships between chunks
2025-02-27 19:17:00 2025-02-27 15:17:00,929 - Time taken to create list chunkids with chunk document: 0.15 seconds
2025-02-27 19:17:00 2025-02-27 15:17:00,932 - Time taken to get the current status of document node: 0.00 seconds
2025-02-27 19:17:00 2025-02-27 15:17:00,932 - Unable to get the status of document node.
2025-02-27 19:17:01 [2025-02-27 15:17:01 +0000] [13] [ERROR] Exception in ASGI application
2025-02-27 19:17:01 Traceback (most recent call last):
2025-02-27 19:17:01   File "/code/score.py", line 234, in extract_knowledge_graph_from_file
2025-02-27 19:17:01     uri_latency, result = await extract_graph_from_file_Wikipedia(uri, userName, password, database, model, wiki_query, language, file_name, allowedNodes, allowedRelationship, token_chunk_size, chunk_overlap, chunks_to_combine, retry_condition, additional_instructions)
2025-02-27 19:17:01   File "/code/src/main.py", line 278, in extract_graph_from_file_Wikipedia
2025-02-27 19:17:01     return await processing_source(uri, userName, password, database, model, file_name, pages, allowedNodes, allowedRelationship, token_chunk_size, chunk_overlap, chunks_to_combine, additional_instructions=additional_instructions)
2025-02-27 19:17:01   File "/code/src/main.py", line 458, in processing_source
2025-02-27 19:17:01     raise LLMGraphBuilderException(error_message)
2025-02-27 19:17:01 src.shared.llm_graph_builder_exception.LLMGraphBuilderException: Unable to get the status of document node.

[abdullah91111]

For more context, I checked the status of the document node "Albert_Einstein" under properties, and the status is "New," so I don't understand why the status isn't passed in /extract endpoint.

[kartikpersistent]

@praveshkumar1988

[praveshkumar1988]

Hi @abdullah91111

You need pass parameter wikiquery and filename are same because in the extraction WikipediaLoader takes query string to search instead of wiki URL. Params looks like..

wiki_query: Albert_Einstein
source_type: Wikipedia
file_name: Albert_Einstein
token_chunk_size: 200
chunk_overlap: 20
chunks_to_combine: 1