0

So I have basic question which I did not get answered by reading the documentation.

  1. If I want to do text classification (sentiment) about lets say an Articel with a topic. I already have the plain text without the html stuff through python libraries. Is it better to make an analysis about each sentence and then combine the results or just pass the whole text as a string and have already combined result (I have already tried the whole text option with flair and it worked quite well I guess).

  2. The next thing would be how you could check if the sentences are about the asked topic and how to check it.

If you could give me some guidelines or hints how to approach these problems I would be happy.

furas
  • 134,197
  • 12
  • 106
  • 148
sabrov
  • 1
  • 1
  • If the goal is to classify the documents, then normally there's no reason to work at the level of sentences first. For question the standard method would be to train a classification model for this task, but this requires some annotated data. – Erwan Jul 28 '22 at 17:52
  • I think this question(s) better match to similar portal [DataScience](https://datascience.stackexchange.com/) – furas Jul 29 '22 at 10:24

0 Answers0