[FEATURE] Ability to execute arbitrary Cypher queries

[tomasonjo]

What is the use case?

Your incremental processing flows have a lot of potential. For example, one cool flow would be to implement MS GraphRAG with it. https://github.com/microsoft/graphrag.

However for downstream steps like entity summarization, I need to be able to execute a cypher query that fetches all nodes that have been just updated to do some summarization which is hard.

Describe the solution you'd like

Instead of spinning up a separate driver for the retrieval, it would be cool to expose such a method for running arbitrary queries directly in cocoindex.

Additional context

One would have to be careful what to run through abstractions and what directly through cypher queries, but it empowers the user a lot

---

❤️ Contributors, please refer to 📙Contributing Guide.
Unless the PR can be sent immediately (e.g. just a few lines of code), we recommend you to leave a comment on the issue like I'm working on it or Can I work on this issue? to avoid duplicating work. Our Discord server is always open and friendly.

[badmonster0]

Make sense! @georgeh0 thoughts on this with https://cocoindex.io/docs/query ?

[badmonster0]

really appreciate your suggestions @tomasonjo we love it!

[georgeh0]

@tomasonjo To clarify if I understand correctly, the requirement here is running Cypher query as a transformation step based on output of previous transformations? Such as

• Step 1: Extract data from source, transform, write into nodes/relationships in Neo4j (this is what CocoIndex can do today)
• Step 2: Run Cypher query from the Neo4j produced by step 1, and create other types of nodes / relationships on top of it (this is what the PR is asking)

If that's the case, this multi-step transformation can be more naturally supported in our next major version.

[tomasonjo]

For example, with ms GraphRAG

1. Extract nodes and rels from docs ( can be done )
2. For all new nodes, check if they already exist in the db and fetch some info from them
3. Run graph algorithms ( so completely independant )