NLP | Location Tags Extraction

Different kind of ChunkParserI subclass can be used to identify the LOCATION chunks. As it uses the gazetteers corpus to identify location words. The gazetteers corpus is a WordListCorpusReader class that contains the following location words:

• Country names
• U.S. states and abbreviations
• Mexican states
• Major U.S. cities
• Canadian provinces

LocationChunker class looking for words that are found in the gazetteers corpus by iterating over a tagged sentence. It creates a LOCATION chunk using IOB tags when it finds one or more location words. The IOB LOCATION tags are produced in the iob_locations() and the parse() method converts the IOB tags to Tree.

Code #1 : LocationChunker class

from nltk.chunk import ChunkParserI
from nltk.chunk.util import conlltags2tree
from nltk.corpus import gazetteers

class LocationChunker(ChunkParserI):
    def __init__(self):
        self.locations = set(gazetteers.words())
        self.lookahead = 0
        for loc in self.locations:
            nwords = loc.count(' ')
        if nwords > self.lookahead:
            self.lookahead = nwords

Code #2 : iob_locations() method

def iob_locations(self, tagged_sent):

    i = 0
    l = len(tagged_sent)
    inside = False

    while i < l:
        word, tag = tagged_sent[i]
        j = i + 1
        k = j + self.lookahead
        nextwords, nexttags = [], []
        loc = False

    while j < k:
        if ' '.join([word] + nextwords) in self.locations:
            if inside:
                yield word, tag, 'I-LOCATION'
            else:
                yield word, tag, 'B-LOCATION'
            for nword, ntag in zip(nextwords, nexttags):
                yield nword, ntag, 'I-LOCATION'
                loc, inside = True, True
                i = j
                break

        if j < l:
            nextword, nexttag = tagged_sent[j]
            nextwords.append(nextword)
            nexttags.append(nexttag)
            j += 1
        else:
            break
        if not loc:
            inside = False
            i += 1
            yield word, tag, 'O'

    def parse(self, tagged_sent):
        iobs = self.iob_locations(tagged_sent)
        return conlltags2tree(iobs)

Code #3 : use the LocationChunker class to parse the sentence

from nltk.chunk import ChunkParserI
from chunkers import sub_leaves
from chunkers import LocationChunker

t = loc.parse([('San', 'NNP'), ('Francisco', 'NNP'),
               ('CA', 'NNP'), ('is', 'BE'), ('cold', 'JJ'),
               ('compared', 'VBD'), ('to', 'TO'), ('San', 'NNP'),
               ('Jose', 'NNP'), ('CA', 'NNP')])

print ("Location : \n", sub_leaves(t, 'LOCATION'))

Output :

Location : 
[[('San', 'NNP'), ('Francisco', 'NNP'), ('CA', 'NNP')], 
[('San', 'NNP'), ('Jose', 'NNP'), ('CA', 'NNP')]]