0
$\begingroup$

I am working on a entity extraction task and I am using Stanford CoreNLP NER. Here, I want to detect entities of type "Animal", "Building", "Imagery", etc., which are not covered in Stanford CoreNLP entity extraction method. Also, I want to detect organization entities which are not included in CoreNLP entity model. I have the data about the organizations with me.

How can I do this?

$\endgroup$
  • $\begingroup$ Welcome to the site! See this question. $\endgroup$ – Emre Mar 27 '18 at 0:18
  • $\begingroup$ Thanks. This works well but suffers from a problem. As this is a dictionary based NER, it does not do the entity disambiguation. For eg, buffalo city can be recognized as the animal buffalo. Currently, I am looking for entity disambiguation techniques. $\endgroup$ – Swastik Roy Apr 16 '18 at 22:07
0
$\begingroup$

There are multiple ways to extract NER. Primarily, based on the construction of statements NER was extracted with the use of POS tags. But overtime with the change of how information was being conveyed, there has been a migration from traditional methods to learning methods. Currently, take a look at sequence to sequence tagging for NER detection. If you have the appropriate dataset, you have the capability to extract anything you consider as NER.

$\endgroup$
  • $\begingroup$ Thank @Nischal, It seems the sequence to sequence tagging needs a lot of training data. Currently, I am focusing on unsupervised or you can say ontology based entity recognition, where I use the entites from public knowledge bases like DbPedia to extract entities. $\endgroup$ – Swastik Roy Apr 16 '18 at 22:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.