Closed PrashantRanjan09 closed 6 years ago
@PrashantRanjan09 can you point us to the referred paper?
It must be this one. https://arxiv.org/abs/1709.04820
Hi @PrashantRanjan09,
A standard approach to improve named entity recognition with continuous word representations is to use word embeddings, trained on large amount of unlabeled data, as feature in the NER system. Examples of this approach can be found in the following papers (in addition to the one from @sughodke's comment):
This general idea was previously explored with discrete word representation (based on word clustering) instead of word embeddings:
Best, Edouard.
I would recommend this paper also which is the foundation of polyglot ner
any code available for https://arxiv.org/abs/1709.04820?
I read that using fastText embeddings can improve the NER. Can you please tell me how can we use fastText embeddings to get the NERs?