LLMs for Schema Generation

I've been playing around with a lot of extraction tasks lately. Often times, i'm reading through information to get a general feel for the schema that would properly extract the 'entities' from any particular unstructured data set. This morning, I had the idea to see how ChatGPT performed on a particular youtube transcript, seems very bloated at the moment. I'm prematurely sharing this, as I wanted to solicit feedback from the group around usage of LLMs for structuring unstructured data.

Has anyone started messing around with this yet?

10 comments