NLP | Splitting and Merging Chunks

SplitRule class : It splits a chunk based on the specified split pattern for the purpose. It is specified like <NN.*>}{<.*> i.e. two opposing curly braces surrounded by a pattern on either side.

MergeRule class : It merges two chunks together based on the ending of the first chunk and the beginning of the second chunk. It is specified like <NN.*>{}<.*> i.e. curly braces facing each other.

Example of how the steps are performed

• Starting with the sentence tree.
  
• Chunking complete sentence.
  
• Chunks are split into multiple chunks.
  
• Chunk with a determiner is split into separate chunks.
  
• Chunks ending with a noun are merged with the next chunk.
  

Code #1 – Constructing Tree

from nltk.chunk import RegexpParser
chunker = RegexpParser(r'''
NP:
{<DT><.*>*<NN.*>}
<NN.*>}{<.*>
<.*>}{<DT>
<NN.*>{}<NN.*>
''')
sent = [('the', 'DT'), ('sushi', 'NN'), ('roll', 'NN'), ('was', 'VBD'),
        ('filled', 'VBN'), ('with', 'IN'), ('the', 'DT'), ('fish', 'NN')]
chunker.parse(sent)

Output:

Tree('S', [Tree('NP', [('the', 'DT'), ('sushi', 'NN'), ('roll', 'NN')]), 
Tree('NP', [('was', 'VBD'), ('filled', 'VBN'), ('with', 'IN')]), 
Tree('NP', [('the', 'DT'), ('fish', 'NN')])])

Code #2 – Splitting and Merging

# Loading Libraries
from nltk.chunk.regexp import ChunkString, ChunkRule, ChinkRule
from nltk.tree import Tree
from nltk.chunk.regexp import MergeRule, SplitRule

# Chunk String
chunk_string = ChunkString(Tree('S', sent))
print ("Chunk String : ", chunk_string)

# Applying Chunk Rule
ur = ChunkRule('<DT><.*>*<NN.*>', 'chunk determiner to noun')
ur.apply(chunk_string)
print ("\nApplied ChunkRule : ", chunk_string)

# Splitting
sr1 = SplitRule('<NN.*>', '<.*>', 'split after noun')
sr1.apply(chunk_string)
print ("\nSplitting Chunk String : ", chunk_string)


sr2 = SplitRule('<.*>', '<DT>', 'split before determiner')
sr2.apply(chunk_string)
print ("\nFurther Splitting Chunk String : ", chunk_string)

# Merging
mr = MergeRule('<NN.*>', '<NN.*>', 'merge nouns')
mr.apply(chunk_string)
print ("\nMerging Chunk String : ", chunk_string)

# Back to Tree
chunk_string.to_chunkstruct()

Output:

Chunk String :   <DT>  <NN>  <NN>  <VBD>  <VBN>  <IN>  <DT>  <NN> 

Applied ChunkRule :  {<DT>  <NN>  <NN>  <VBD>  <VBN>  <IN>  <DT>  <NN>}

Splitting Chunk String :  {<DT>  <NN>}{<NN>}{<VBD>  <VBN>  <IN>  <DT>  <NN>}

Further Splitting Chunk String :  {<DT>  <NN>}{<NN>}{<VBD>  <VBN>  <IN>}{<DT>  <NN>}

Merging Chunk String :  {<DT>  <NN>  <NN>}{<VBD>  <VBN>  <IN>}{<DT>  <NN>}

Tree('S', [Tree('CHUNK', [('the', 'DT'), ('sushi', 'NN'), ('roll', 'NN')]), 
          Tree('CHUNK', [('was', 'VBD'), ('filled', 'VBN'), ('with', 'IN')]), 
          Tree('CHUNK', [('the', 'DT'), ('fish', 'NN')])])