B
B
beduin012021-01-12 17:01:31
natural language processing
beduin01, 2021-01-12 17:01:31

Do I need NLP?

The language is not critical. The problem is as follows. There are many pieces of text like:
The decision was canceled, Denied, Documents were not submitted correctly, Incorrect number, etc. There are a lot of pieces

and they cannot be reduced to a dictionary. Matching with the pattern of "cancels" and "not [correct | correct]" seems to me not the best option.

I need to understand piece by piece that some kind of negation is going on and set the False flag.

Are there any other options? Do we need pure NLP here? Or are there other resources?

Answer the question

In order to leave comments, you need to log in

2 answer(s)
X
xmoonlight, 2021-01-12
@xmoonlight

If something is specified, then the status of the document must clearly be.
If there is a status, you can do a "subtraction": "not positive" means "negative".
It remains to find all the "positive" (approved) statuses.

S
Sergey, 2021-01-13
@begemot_sun

I don't know if a naive bayes classifier can be called NLP.
But feed this kind of texts to him, let him learn.
Code example: https://github.com/loguntsov/bayes
you can play around with your data.

Didn't find what you were looking for?

Ask your question

Ask a Question

731 491 924 answers to any question