search

sdsc-ordes / kg-llm-interface

Langchain-powered natural language interface to knowledge-graphs.
Apache License 2.0
14 stars 3 forks source link

Rdf ingestion #6

Closed cmdoret closed 1 year ago

cmdoret commented 1 year ago

This PR simplifies the chroma indexing flow and adds an RDF ingestion flow.

  1. aikg/flows/insert_triples.py: Load RDF files (e.g. turtle) to a SPARQL endpoint
  2. aikg/flows/chroma_build.py: SPARQL endpoint -> ChromaDB collection

These scripts are provided for users who want to bring their own data to run the server, one just needs a Chroma server with a collection ready.

caviri commented 1 year ago

Hello @cmdoret I'm trying to check this pull request.

Shall I use the docker compose up for it? I guess in this case I need to have a collection already in the chroma right?

I'm having this error: AttributeError: module 'aikg.config.chroma' has no attribute 'Config'

cmdoret commented 1 year ago

Thanks @caviri I fixed the outdated import paths that caused the error. Yes ultimately we'll need to provided a demo collection for it to work out of the box, but maybe this could be handled in a separate PR?