RAG implementation from an existing knowledge graph

[laurasck14]

Checked other resources

• I added a very descriptive title to this question.
• I searched the LangChain documentation with the integrated search.
• I used the GitHub search to find a similar question and didn't find it.

Commit to Help

• I commit to help with one of those options 👆

Example Code

from langchain_community.graphs import NebulaGraph

graph = NebulaGraph(
    space= "PrimeKG",
    username="root",
    password="nebula",
    address="127.0.0.1",
)

chain = NebulaGraphQAChain.from_llm(
    llm=llm,
    graph=graph,
    verbose=True,
    allow_dangerous_requests=True,
)

Description

I am trying to implement RAG from and EXISTING knowledge graph created outside of Langchain. What I mean is that I already have data for nodes and edges and I saw Langchain already supports creation of them.

All of my nodes have some metadata information but not all of them have an associated text (from a compilation of different sources).

All of the documentation and examples I saw so far are based on documents from which a graph is then created. This would not be needed but I would like to generate embeddings from the nodes that take into account the text/metadata/edges...

My questions:

• is there a way in Langchain to implement a RAG from an existing knowledge graph? if so, what would be the best way?
• can embeddings be generated from nodes to speed up the retrieval? (combine semantic search with the embeddings)

I'm kind of new to NLP and RAG so any help is appreciated! :)

System Info

System Information

OS: Linux
OS Version: #1 SMP Thu Aug 8 17:47:08 UTC 2024
Python Version: 3.12.0 | packaged by conda-forge | (main, Oct 3 2023, 08:43:22) [GCC 12.3.0]

Package Information

langchain_core: 0.3.34
langchain: 0.3.18
langchain_community: 0.3.15
langsmith: 0.3.1
langchain_graphrag: 0.0.9
langchain_huggingface: 0.1.2
langchain_text_splitters: 0.3.6
langchainhub: 0.1.21
langgraph_sdk: 0.1.51