FalkorDB Graph Store

Topics

• Overview
• Install package
• Registering FalkorDB as a graph store
• Creating a FalkorDB graph store

Overview

You can use FalkorDB as a graph store.

Install package

The FalkorDB graph store is contained in a separate contributor package. To install it:

!pip install https://github.com/awslabs/graphrag-toolkit/archive/refs/tags/v3.15.5.zip#subdirectory=lexical-graph-contrib/falkordb

Registering FalkorDB as a graph store

Before creating a FalkorDB graph store, you must register the FalkorDBGraphStoreFactory with the GraphStoreFactory:

from graphrag_toolkit.lexical_graph.storage import GraphStoreFactory
from graphrag_toolkit_contrib.lexical_graph.storage.graph.falkordb import FalkorDBGraphStoreFactory

GraphStoreFactory.register(FalkorDBGraphStoreFactory)

Creating a FalkorDB graph store

You can use the GraphStoreFactory.for_graph_store() static factory method to create an instance of a FalkorDB graph store.

The FalkorDB graph store currently supports semantic-guided search. It does not support traversal-based search.

To create a FalkorDB Cloud graph store, supply a connection string that begins falkordb://, followed by the FalkorDB endpoint:

from graphrag_toolkit.lexical_graph.storage import GraphStoreFactory
from graphrag_toolkit_contrib.lexical_graph.storage.graph.falkordb import FalkorDBGraphStoreFactory

falkordb_connection_info = 'falkordb://your-falkordb-endpoint'

GraphStoreFactory.register(FalkorDBGraphStoreFactory)

with GraphStoreFactory.for_graph_store(falkordb_connection_info) as graph_store:
  ...

You may also need to pass a username and password, and specify whether or not to use SSL:

from graphrag_toolkit.lexical_graph.storage import GraphStoreFactory

falkordb_connection_info = 'falkordb://<your-falkordb-endpoint>'

with GraphStoreFactory.for_graph_store(
      falkordb_connection_info,
      username='<username>',
      password='<password>',
      ssl=True
  ) as graph_store:

    ...

To create a local FalkorDB graph store, supply a connection string that has only falkordb://;

from graphrag_toolkit.lexical_graph.storage import GraphStoreFactory

falkordb_connection_info = 'falkordb://'

with GraphStoreFactory.for_graph_store(falkordb_connection_info) as graph_store:
  ...