You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Placename recognition is an essential technique that pipeline should master.
Social scientists see placename as a piece of important information in commenting about the events.
the model we are using is not doing very well so I tried using NLTK, Standford and Spacy NER.
the results as follows,
% of 1-gram matching
match token number
gold_tokens
1.000.000
177
used_classifier (BERT)
0.610169
108
nltk
0.723164
128
spacy
0.000000
0
standford
0.423729
75
NeuroNER
0.203390
36
The text was updated successfully, but these errors were encountered:
We are now using SoTA Flair's place name model (en-ner-conll03-v0.4) as an extra place name extractor.
This model has a recall of 97% in our test data.
Placename recognition is an essential technique that pipeline should master.
Social scientists see placename as a piece of important information in commenting about the events.
the model we are using is not doing very well so I tried using NLTK, Standford and Spacy NER.
the results as follows,
The text was updated successfully, but these errors were encountered: