R2R icon indicating copy to clipboard operation
R2R copied to clipboard

Knowledge Graph from Entity Extraction

Open taahan0810 opened this issue 1 year ago • 6 comments
trafficstars

Hey, I was thinking of creating a knowledge graph from the extracted entities of the documents. We could first extract the entities and then these entities could be the edges of the documents. This way we can map the documents as knowledge where the edge information is the common entities between documents. How does this sound?

taahan0810 avatar Mar 30 '24 13:03 taahan0810

Sounds amazing, let's start with an entity extraction pipeline that can sit after the ingestion pipeline in the default RAG system?

emrgnt-cmplxty avatar Mar 30 '24 17:03 emrgnt-cmplxty

Sounds good!

taahan0810 avatar Mar 30 '24 17:03 taahan0810

@taahan0810 - I spent a few minutes and whipped up a very RD of the entity extraction pipeline here, thoughts?

emrgnt-cmplxty avatar Mar 30 '24 20:03 emrgnt-cmplxty

This looks good! I was not sure if the Pipeline class import was present in r2r.pipelines so I left a comment.

taahan0810 avatar Mar 30 '24 21:03 taahan0810

this code won't run, it's more or less an early version - do you want to pull it in and polish it and commit some changes?

emrgnt-cmplxty avatar Mar 31 '24 00:03 emrgnt-cmplxty

Yes, on it!

taahan0810 avatar Mar 31 '24 00:03 taahan0810