I am trying to build a NLP app which essentially has to do Named Entity Recognition (NER). I came across GATE. From what i understand it is a framework to build NLP apps. I tested ANNIE, the IE system distributed with GATE but the NER results for my domain is not up-to the expectation. As a matter of fact any NER, like Stanford CoreNLP or NLTK, is not giving me required results. So i decide to tweak the existing systems to get desired result.
Regarding GATE i liked few things:
1. The modularity of components: For example in ANNIE, components like Tokenizer, Gaztteer, Sentence splitter, POS tagger etc can be used independently of each other.
2. Its rule language called JAPE which has a very nice way of writing rules or patterns.
But few things i want to know about GATE are:
1. What are the other major advantages of GATE particularly for NER?
2. How flexible is GATE for adding new components? For example some day if i want to use NLTK's POS tagger inside GATE?
3. If i want to use custom machine learning models with GATE?
4. I am aware that NLP group at University of Sheffield is involved in GATE, but i want to know how active is GATE's community and how active is the support for GATE?
5. Can GATE be used for commercial software?
Keen to here suggestions from people who have actually used GATE