I posted this request but it was linked to an anonymous account, so I am reposting.
My PhD research was in medical term recognition using the NIH medical ontology database. I was able to develop a relatively simple neural network that achieved significant levels of precision for recognizing terms, with an F1 score of 0.97.
Has anyone done medical term recognition, where the terms are in the NIH ontologies, using Hugging Face? I’d like to compare Hugging Face to what I was able to produce.
In my research I used a hand-annotated text, available from the NIH, as ground truth. I’d like to see how Hugging Face performs against this same ground truth.