Using Stanford CoreNLP Python Parser for specific output

Question

I'm using SCP to get the parse CFG tree for English sentences.

from corenlp import *
corenlp = StanfordCoreNLP()
corenlp.parse("Every cat loves a dog")

My expected output is a tree like this:

(S (NP (DET Every) (NN cat)) (VP (VT loves) (NP (DET a) (NN dog))))

But what i got is:

(ROOT (S (NP (DT Every) (NN cat)) (VP (VBZ loves) (NP (DT a) (NN dog)))))

How to change the POS tag as expected and remove the ROOT node?

Thanks

Answer 1

You can use nltk.tree module from NLTK.

from nltk.tree import *

def traverse(t):
    try:
        # Replace Labels
        if t.label() == "DT":
            t.set_label("DET")
        elif t.label() == "VBZ":
            t.set_label("VT")   
    except AttributeError:
        return

    for child in t:
        traverse(child)

output_tree= "(ROOT (S (NP (DT Every) (NN cat)) (VP (VBZ loves) (NP (DT a) (NN dog)))))"
tree = ParentedTree.fromstring(output_tree)

# Remove ROOT Element
if tree.label() == "ROOT":  
    tree = tree[0]

traverse(tree)
print tree  
# (S (NP (DET Every) (NN cat)) (VP (VT loves) (NP (DET a) (NN dog))))