You are viewing a plain text version of this content. The canonical link for it is here.

Posted to dev@opennlp.apache.org by sri krishna <kr...@yahoo.com> on 2011/12/17 04:10:59 UTC

Name entity recognization(NER)

1) After running the test set on the created model is there a way where we can get the recognized entity mapping with the nearest actual training tuple or probabilities with the 

corresponding training set entries, as i have most of the entities in my train set.

2) It seems that CLI and API are generating different results for same model and test set.

Re: Name entity recognization(NER)

Posted by Jörn Kottmann <ko...@gmail.com>.

On 12/17/11 4:10 AM, sri krishna wrote:
> It seems that CLI and API are generating different results for same model and test set

This might be caused by the following:
- Different tokenization
- Different sentence bounds, cli has one sentence per line
- You don't clear the adaptive data, in the cli an empty line clears it, 
in the api
    you need to call clearAdaptiveData on the name finder.

Jörn