Natural Language Processing Package

Question

I have started working on a project which requires Natural Language Processing. We have do the spell checking as well as mapping sentences to phrases and their synonyms. I first thought of using GATE but i am confused on what to use? I found an interesting post here which got me even more confused.

http://lordpimpington.com/codespeaks/drupal-5.1/?q=node/5

Please help me decide on what suits my purpose the best. I am working a web application which will us this NLP tool as a service.

Mottled

There's no such thing, in a sense, as 'NLP'. It would be better if you would be really specific as to the tasks you need to perform. – Panoply

See what i basically need is any basic architecture/tool which would help me work on setting up machine learning systems. The sentences which come in as input need to be mapped to certain phrases/synonyms of phrases which may be present in the sentence. Which sentiment of the sentence maps to which of the phrases set up by me. – Mottled

Looks like a duplicate of #2062381 to me. – Lorrimor

Answer 1

You didn't really give much info, but try this: http://www.nltk.org/

I don't think NLTK does spell checking (I could be wrong on this), but it can do parts of speech tagging for text input.

For finding/matching synonyms you could use something like WordNet http://wordnet.princeton.edu/

If you're doing something really domain specific: I would recommend coming up with your own ontology for domain specific terms.

Lineal

There are multiple domains for which the analysis has to be done. What would you suggest in such a case. For a basic set up what can i use. Something like a GATE.. How useful can it be? – Mottled

It depends how closely they match a given ontology (such as WordNet). If you are using most terms in the same way that a pre-defined ontology is, then you don't need to roll your own. If you are using the same terms in different ways then you may need to create a new one. If you have multiple domains that use terms in different ways, then you may need to create more than one. – Lineal

Answer 2

If you are using Python you can develop a spell checker with Python Enchant. NLTK is good for developing Sentiment Analysis system too. I have some prototypes of the same too

Jaggu

Unprecedented

Answer 3

If you are using deep learning based models, and if you have sufficient data, you can implement task specific models for any purpose. With the development of deep leaning based languages models, you can used word embedding based models with lexicon resources to obtain synonyms and antonyms. You can also follow the links below to obtain more resources.

• https://stanfordnlp.github.io/CoreNLP/
• https://www.nltk.org/
• https://wordnet.princeton.edu/

Tragicomedy