-1

I want to parse out all person name from the input text data. I already tried it by using Stanford Java NER library but it does not give accurate answers for indian Names. Is it possible to create customised classifiers? or is there any other method to parse names using other NLP techniques with maximum accuracy.

EDIT: I gave random text as input containing many person names to the GUI application of Stanford Named Entity Recognizer version 3.5.1 But it is very less accurate. It does not recognize most of the names.

ROHIT IHARE
  • 46
  • 4
  • Please do not ask generic questions. Tell us what you have tried and how it failed. Which approach at least gave some basic results. People here then can suggest improvements. – rishi Feb 25 '15 at 11:15

1 Answers1

0

Stanford NLP documentation tells that you can train your own NER Model. Here's the link to the [FAQ][1][1]: http://nlp.stanford.edu/software/crf-faq.shtml. You can use a moderately rich text with person names and use it in training.

There are other tools (for example, GATE) having a list of person names, which is modifiable.

Also, look around in NLTK, LingPipe, FreeLing etc. which are other popular NLP libraries.

pnv
  • 1,437
  • 3
  • 23
  • 52