Graph RAG

Graph RAG provides retrievers that combine unstructured similarity-search on vectors and structured traversal of metadata properties. These retrievers are implemented using the metadata search functionality of existing vector stores, allowing you to traverse your existing vector store!

• Link based on existing metadata

  ---

  Use existing metadata fields without additional processing. Retrieve more from your existing vector store!

  Get started

• Change links on demand

  ---

  Edges can be specified on-the-fly, allowing different relationships to be traversed based on the question.

  Edges

• Pluggable Traversal Strategies

  ---

  Use built-in traversal strategies like Eager or MMR, or define your own logic to select which nodes to explore.

  Strategies

• Broad compatibility

  ---

  Adapters are available for a variety of vector stores with support for additional stores easily added.

  Adapters

Example: LangChain Retriever combining Vector and Graph traversal

from langchain_graph_retriever import GraphRetriever
from graph_retriever.edges import Id

retriever = GraphRetriever(
    store = store,
    edges = [("mentions", Id()), ("entities", "entities")], # (1)!
)

retriever.invoke("where is Santa Clara?")

1. edges configures traversing from a node to other nodes listed in the metadata["mentions"] field (to the corresponding id) and to other nodes with overlapping metadata["entities"].

See Examples for more complete examples.