AI applicationsAI Toolkit and integrationsLightRAG

LightRAG

LightRAG is a retrieval-augmented generation (RAG) framework that extracts entities and relationships from your documents into a knowledge graph and retrieves from it to answer questions. It supports Memgraph as its graph storage. Unstructured2Graph can use LightRAG as its extraction backend to build knowledge graphs in Memgraph.

Installation

Install LightRAG:

pip install lightrag-hku

Environment setup

Start Memgraph, for example in Docker:

docker run -p 7687:7687 memgraph/memgraph:latest

Set the connection details. These are the defaults for a local Memgraph without authentication:

export MEMGRAPH_URI="bolt://localhost:7687"
export MEMGRAPH_USERNAME=""
export MEMGRAPH_PASSWORD=""
export MEMGRAPH_DATABASE="memgraph"

The example uses OpenAI, so also set OPENAI_API_KEY.

Basic usage

This script inserts text, builds the graph in Memgraph and runs a hybrid query. graph_storage="MemgraphStorage" tells LightRAG to store the graph in Memgraph:

import os
import asyncio
from lightrag import LightRAG, QueryParam
from lightrag.llm.openai import gpt_4o_mini_complete, gpt_4o_complete, openai_embed
from lightrag.kg.shared_storage import initialize_pipeline_status
from lightrag.utils import setup_logger
 
setup_logger("lightrag", level="INFO")
 
WORKING_DIR = "./rag_storage"
if not os.path.exists(WORKING_DIR):
    os.mkdir(WORKING_DIR)
 
async def initialize_rag():
    rag = LightRAG(
        working_dir=WORKING_DIR,
        embedding_func=openai_embed,
        llm_model_func=gpt_4o_mini_complete,
        graph_storage="MemgraphStorage",  # store the graph in Memgraph
    )
    # Both initialization calls are required
    await rag.initialize_storages()
    await initialize_pipeline_status()
    return rag
 
async def main():
    rag = None
    try:
        rag = await initialize_rag()
        rag.insert("Your text")
 
        print(
          await rag.query(
              "What are the top themes in this story?",
              param=QueryParam(mode="hybrid")
          )
        )
 
    except Exception as e:
        print(f"An error occurred: {e}")
    finally:
        if rag:
            await rag.finalize_storages()
 
if __name__ == "__main__":
    asyncio.run(main())

Configuration

Workspace isolation: to keep the data of several LightRAG instances apart, set the MEMGRAPH_WORKSPACE environment variable:

export MEMGRAPH_WORKSPACE="project_a"

Configuration file: instead of environment variables, you can set the connection in a config.ini file:

[memgraph]
uri = bolt://localhost:7687
username =
password =
database = memgraph

For more, see the LightRAG integration in the Memgraph AI Toolkit, the Jupyter notebook example and the LightRAG repository.

LlamaIndexCognee