Open kwalcock opened 1 week ago
Fwiw, the latest processors labels both these instances as person.
@MihaiSurdeanu, thanks for checking. I'll be working on the part that converts the collected documents into a database sometime soon and it may be worth revisiting a few issues like this. We run processors once already as part of eidos for causality and recently it has been running again for lemma and NE extraction. We could have this named entity information for free if the pipeline was rearranged a little to reuse processors output. There wouldn't be a need to run spacy.
I don't know whether we can do anything about it, but Nana here probably refers to a person, Nana Kwame Opoku, rather than location 12168625.
We are using spacy for NER in the location code. Can we do any better with processors?