search

SciPhi-AI / R2R

The most advanced Retrieval-Augmented Generation (RAG) system, containerized and RESTful
https://r2r-docs.sciphi.ai/
MIT License
3.64k stars 270 forks source link

Integration with neo4j vector index #601

Open jexp opened 4 months ago

jexp commented 4 months ago

It's great that neo4j is already your KG provider, not sure if you're aware of but Neo4j added vector indexing and search (HNSW) last summer. So you can do everything with one provider and don't need to set up two.

LMK if you have any questions, we're happy to help with a PR.

It's also used in our LangChain, LlamaIndex, SpringAI, Haystack, Semantic Kernel, LangChain4j integrations.

See:

https://neo4j.com/labs/genai-ecosystem/vector-search/ https://neo4j.com/docs/cypher-manual/current/indexes/semantic-indexes/vector-indexes/ https://neo4j.com/developer-blog/neo4j-langchain-vector-index-implementation/

schoi80 commented 3 months ago

This implementation for postgres looks fairly straight-forward. It would be great to see VectorDatabaseProvider implementation for neo4j support here as well. I might take a stab at this (or happy to contribute if @jexp has already started 😉)

emrgnt-cmplxty commented 2 months ago

Circling back on this, the codebase is more stable now so we'd love to see this integrated! It would be best to start work on this with the major revisions we are rolling into the V3 - https://github.com/SciPhi-AI/R2R/pull/877