NLP | Flattening Deep Tree

Some of the corpora that we use are often deep trees of nested phrases. But working on such deep trees is a tedious job for training the chunker. As IOB tag parsing is not designed for nested chunks. So, in order to use these trees for chunker training, we must flatten them.
Well, POS (part of Speech) are actually part of the tree structure instead of being in the word. These are used with Tree.pos() method, designed specifically for combining words with preterminal Tree labels such as part-of-speech tags.

Code #1 : Class for flattening the deep tree

from nltk.tree import Tree

def flatten_childtrees(trees):
    children = []

    for t in trees:
        if t.height() < 3:
            children.extend(t.pos())

        elif t.height() == 3:
            children.append(Tree(t.label(), t.pos()))

        else:
            children.extend(
                    flatten_childtrees())

    return children


def flatten_deeptree(tree):
    return Tree(tree.label(),
                flatten_childtrees())

Code #2 : Evaluating flatten_deeptree()

from nltk.corpus import treebank
from transforms import flatten_deeptree

print ("Deep Tree : \n", treebank.parsed_sents()[0])

print ("\nFlattened Tree : \n",
       flatten_deeptree(treebank.parsed_sents()[0]))

Output :

Deep Tree : 
 (S
  (NP-SBJ
    (NP (NNP Pierre) (NNP Vinken))
    (,, )
    (ADJP (NP (CD 61) (NNS years)) (JJ old))
    (,, ))
  (VP
    (MD will)
    (VP
      (VB join)
      (NP (DT the) (NN board))
      (PP-CLR (IN as) (NP (DT a) (JJ nonexecutive) (NN director)))
      (NP-TMP (NNP Nov.) (CD 29))))
  (. .))

Flattened Tree : 
Tree('S', [Tree('NP', [('Pierre', 'NNP'), ('Vinken', 'NNP')]), (', ',
', '), Tree('NP', [('61', 'CD'), ('years', 'NNS')]), ('old', 'JJ'),
(', ', ', '), ('will', 'MD'), ('join', 'VB'), Tree('NP', [('the',
'DT'), ('board', 'NN')]), ('as', 'IN'), Tree('NP', [('a', 'DT'),
('nonexecutive', 'JJ'), ('director', 'NN')]), Tree('NP-TMP', [('Nov.',
'NNP'), ('29', 'CD')]), ('.', '.')])

The result is a much flatter Tree that only includes NP phrases. Words that are not part of an NP phrase are separated

How it works ?

• flatten_deeptree() : returns a new Tree from the given tree by calling flatten_childtrees() on each of the given tree’s children.
• flatten_childtrees() : Recursively drills down into the Tree until it finds child trees whose height() is equal to or less than 3.

Code #3 : height()

from nltk.corpus import treebank
from transforms import flatten_deeptree

from nltk.tree import Tree

print ("Height : ",
       Tree('NNP', ['Pierre']).height())

print ("\nHeight : ", Tree(
        'NP', [Tree('NNP', ['Pierre']),
                    Tree('NNP', ['Vinken'])]). height())

Output :

Height : 2

Height : 3