How to map Extracted Data onto onvoc

[jdkent]

1. get labels (and definitions) from the papers (extraction LLM)
2. get embeddings of labels + definitions
3. expand relevant onvoc terms (mapping onto different ontologies, either manually or semi-assisted)
4. get embeddings of onvoc terms
5. take the embeddings of labels + definitions (along with the text) and use a topic model (like Topeax) with seed words/embeddings from onvoc for each term I'm interested in.
6. map the papers onto the topics and then combine all the created topics (not a part of onvoc) as potential search terms.