If I have a document with many paragraphs of text, would using a custom NER model be suitable to identify a sentence as a recognized entity? The desired sentences will be similar in semantic structure--but not similar enough to use RegEx!
Asked
Active
Viewed 28 times
1
-
NER models usually detects compound words at best not entire sentence. If we talk about sentence similarity, NER is not suitable approach. – msuzen Jun 29 '21 at 14:37
-
This sounds like an [XY problem](https://xyproblem.info)—what are you really trying to do? – Arya McCarthy Jun 29 '21 at 20:41
-
@AryaMcCarthy I need to parse a fairly large document for sentences that convey certain meaning. This appears to be a classification problem, and I thought NER could be suitable because NER models can also identify spans as well as other entities. I was curious if traditional entities could be expanded to capture broader ideas. – Jason p Jun 29 '21 at 21:41