Is it possible in spacy to detect paragraphs using NER? #10323
-
So i'm having some problems here in detecting long sentences. One example is my document is around 30k characters long and my label is around 1 paragraph long. Is it possible to detect it properly? |
Beta Was this translation helpful? Give feedback.
Replies: 2 comments
-
Example is |
Beta Was this translation helpful? Give feedback.
-
Hi @hacksider , you can try using the SpanCategorizer and see if it fits your use-case. Just note that this feature is currently experimental, but it's designed to work on long contiguous spans. For this, you need to store your entities in |
Beta Was this translation helpful? Give feedback.
Hi @hacksider , you can try using the SpanCategorizer and see if it fits your use-case. Just note that this feature is currently experimental, but it's designed to work on long contiguous spans. For this, you need to store your entities in
doc.spans[spans_key]
instead ofdoc.ents
, so you might have to process your texts a bit (cf. storing via a SpanGroup, SpanCat vs. NER).