NLP | Customization Using Tagged Corpus Reader

How we can use Tagged Corpus Reader ?

• Customizing word tokenizer
• Customizing sentence tokenizer
• Customizing paragraph block reader
• Customizing tag separator
• Converting tags to a universal tagset

Code #1 : Customizing word tokenizer

Python3

# Loading the libraries
from nltk.tokenize import SpaceTokenizer
from nltk.corpus.reader import TaggedCorpusReader

x = TaggedCorpusReader('.', r'.*\.pos',
                       word_tokenizer = SpaceTokenizer())

x.words()

Output : 

['The', 'expense', 'and', 'time', 'involved', 'are', ...]

Code #2 : For sentence

Python3

# Loading the libraries
from nltk.tokenize import LineTokenizer
from nltk.corpus.reader import TaggedCorpusReader

x = TaggedCorpusReader('.', r'.*\.pos',
                       sent_tokenizer = LineTokenizer())

x.sents()

Output : 

[['The', 'expense', 'and', 'time', 'involved', 'are', 'astronomical', '.']]

Customizing paragraph

• Assume paragraphs to be split by blank lines
• Done with the para_block_reader function, which is nltk.corpus.reader.util.read_blankline_block
• Number of other block reader are present in nltk.corpus.reader.util, whose purpose is to read blocks of text from a stream.

Customizing Tag separator

• If ‘/’ is not used as the word/tag separator, one can pass an alternative string to TaggedCorpusReader for sep.
• Default is sep=’/’, but if one wants to split words and tags with ‘|’, such as ‘word|tag’, then sep=’|’ is passed in .

Converting tags to a universal tagset
Tagset : It is a list of POS tags used by one or more corpora. 
Universal Tagset : It is a simplified and condensed tagset composed of only 12 part-of-speech tags
Code #3 : map corpus tags to the universal tagset

Python3

from nltk.corpus.reader import TaggedCorpusReader

x = TaggedCorpusReader('.', r'.*\.pos', tagset ='en-brown')
x.tagged_words(tagset ='universal')

Output : 

[('The', 'DET'), ('expense', 'NOUN'), ('and', 'CONJ'), ...] 

Code #4 : map corpus tags to the universal tagset

Python3

from nltk.corpus.reader import TaggedCorpusReader
from nltk.corpus import treebank

treebank.tagged_words()

treebank.tagged_words(tagset ='universal')

treebank.tagged_words(tagset ='brown')

Output : 

[('Pierre', 'NNP'), ('Vinken', 'NNP'), (', ', ', '), ...]

[('Pierre', 'NOUN'), ('Vinken', 'NOUN'), (', ', '.'), …]

[('Pierre', 'UNK'), ('Vinken', 'UNK'), (', ', 'UNK'), ...]