0

I'm currently preparing a NER Task with Flair and I'm looking for some information about metrics used for NER task.

What are the most used metrics and how to interpretate them ?

Nicolas Roche
  • 105
  • 1
  • 1
  • 14
  • Would the [paper behind Flair](http://alanakbik.github.io/papers/coling2018.pdf) be anything close to an answer, specifically the evaluation section? Also I think this might be slightly off-topic for SO, since you're trying to optimize towards specific metrics depending on your respective task setup and environment (academic/industry)... This can attract strongly opinionated answers – dennlinger Jan 13 '20 at 10:07

1 Answers1

0

I would suggest checking sklearn-crfsuite's documentationif you want out of the box implementation.

For interpretation consider your NER system as a multiclass classification system.

ashutosh singh
  • 511
  • 3
  • 15