A
A
andrkkk2020-03-30 12:34:21
Node.js
andrkkk, 2020-03-30 12:34:21

How to handle words written in different ways?

For example, a car has a power steering, in the text it can be written in full or, for example, power steering, how to extract such words from the text and classify them?
I only see creating a small database of spellings for each word, is that correct?

Answer the question

In order to leave comments, you need to log in

2 answer(s)
V
Vladimir, 2020-03-30
@andrkkk

Yes, I think it’s quite, but ideally, establish rules for writing something like this - either briefly or completely

X
xmoonlight, 2020-03-30
@xmoonlight

You can look at the first letters (and the first few letters, and choose the best option by matches) in all the roots in a row of words, taking into account the prepositions between them with a probability of 50% (check with and without prepositions, and choose the best match).
This will cover most cases.
And you can also make your own dictionary for several special cases: the more classifiers, the better.

Didn't find what you were looking for?

Ask your question

Ask a Question

731 491 924 answers to any question