How to extract the context from the text and understand what it refers to?

Question

Machine learning

Hello everyone, I've started doing NLP. I would like to know how to solve this problem. I have a dataset of people's appeals, for example, to a clinic. By type: My head hurts, I feel sick or I fell down and my arm hurts. I want to find a way to refer these calls to the right doctor. For example, the hand hurts fell - chiropractor 80%. There is an option to isolate for a start the context of how dizzy and so on. And then through the National Assembly to correlate these phrases with doctors.

I don't know much about this part of ML. So it would be very helpful to hear ideas on this.

Answer 1

If it is true that you "I don't understand this particular part of ML", i.e. - in other sections you understand well - then I give a tip. The task refers to the tasks of classification. First, text vectorization is carried out, then generally standard supervised training procedures are applied. The task is perhaps one of the simplest in the field of NLP. For texts - including both short messages and full-text ones - it has been solved for a long time and quite successfully. At one time, it seems Reuters boasted that it was able to replace about 100 people classifying texts into sections with one NLP program. So they also have a multi-class classification.
There is more than enough information on the problem in general, and on this particular task on the Internet. Yes, and in books it has already been described, up to implementations, for example
S. Bird, E. Klein, E. Loper. Natural Language Processing with Python
B. Bengfort, R. Bilbro, T. Ojeda. Applied Text Analysis with Python
and more:
https://medium.com/sciforce/top-10-books-on-nlp-an...
https://towardsdatascience.com/top-nlp-books-to-re. ..
https://www.tableau.com/learn/articles/natural-lan...
With such an abundance and details of the presentation of information, in general, it is not very clear what kind of help you need. Just don’t “tell me how to solve this problem”, but “I solve this problem with this method and I don’t understand this question here.” Ask - we can help.