Equivalent of nltk.RegexpParser

[OmarJay1]

Hello, is there anything in wink-nlp that is equivalent to nltk.RegexpParser?

I'm trying to do something like the following, and the one thing I don't readily see is being able to classify grammar sequences.

Thanks.

 tokens = nltk.word_tokenize(title)
        tagged = nltk.pos_tag(tokens)
        grammar =   """  
                    LOCATION: {<IN><NNP>+<,|IN><NNP>+} 
                    PROPER: {<NNP|NNPS>+}
                    NOUN: {<NN>+}
                    """       

        chunker = nltk.RegexpParser(grammar)
        result = chunker.parse(tagged)

[rachnachakraborty]

Hey @OmarJay1

Equivalent of nltk.RegexpParser is not available in winkNLP as of now. It's in our roadmap. You can explore the custom entities extraction which can serve the purpose for the type of use case you have mentioned.

A sample code is attached for your reference.

const winkNLP = require( 'wink-nlp' );
const model = require( 'wink-eng-lite-web-model' );
const nlp = winkNLP( model );
const its = nlp.its;

const patterns = [
  { name: 'LOCATION', patterns: [ '[in|from|to] [PROPN] [PROPN] [of] [PROPN]' ], mark: [1, 4] },
  { name: 'LOCATION', patterns: [ '[in|from|to] [PROPN]' ], mark: [1, 1] },
  { name: 'LOCATION', patterns: [ '[in|from|to] [PROPN] [PROPN]' ], mark: [1, 2] },
  { name: 'LOCATION', patterns: [ '[in|from|to] [PROPN] [PROPN] [PROPN]' ], mark: [1, 3] },
  { name: 'PROPER', patterns: [ '[PROPN] [|PROPN] [|PROPN] [|PROPN]' ] },
  { name: 'NOUN_PHRASE', patterns: [ '[|DET] [|ADJ] [NOUN] [|NOUN] [|NOUN]' ] }
];
nlp.learnCustomEntities( patterns );
const text = 'Narendra Modi was born in India. Joe Biden was born in United States of America. I travelled from South America  to United Kingdom. The enduring strength of our values.';
const doc = nlp.readDoc( text );
console.log( doc.customEntities().out( its.detail ) );

Output

Array (8 items)
0: Object {value: "Narendra Modi", type: "PROPER"}
1: Object {value: "India", type: "LOCATION"}
2: Object {value: "Joe Biden", type: "PROPER"}
3: Object {value: "United States of America", type: "LOCATION"}
4: Object {value: "South America", type: "LOCATION"}
5: Object {value: "United Kingdom", type: "LOCATION"}
6: Object {value: "The enduring strength", type: "NOUN_PHRASE"}
7: Object {value: "values", type: "NOUN_PHRASE"}
Array Prototype

You can try it on runkit with Node v 16 and above.

Best,
Rachna

[OmarJay1]

Hi @rachnachakraborty, the code you provided works really well.

Thank you so much,

Omar