benepar-nx 0.0.1

NetworkX graph from benepar constituency parse

benepar_nx

NetworkX graph from Berkeley neural parser constituency parse

Getting started , or see demo.ipynb

FIXME: get benepar

pip install benepar && \
python -c "import benepar; benepar.download('benepar_en3')"

and maybe spaCy model

python -m spacy download en_core_web_md

Add benepar to the pipeline

import spacy
md_benepar = spacy.load('en_core_web_md')
import benepar # for add_pipe('benepar'...

md_benepar.add_pipe('benepar', config={'model': 'benepar_en3'})

parse sentences:

doc = md_benepar('In the fourth quarter, we closed on the transformative acquisitions of Larry H. Miller and Total Care Auto, powered by Landcar, Kahlo Chrysler Jeep Dodge, Arapahoe Hyundai-Genesis and the Stevinson Automotive Group, representing approximately $6.6 billion in annualized revenue.')

and create graph for the sentence:

from benepar_nx.constituency_parse import create_constituency_parse_graph

s = list(doc.sents)[0]
G = create_constituency_parse_graph(s)

Draw constituency parse graph

Get graphviz , GNU sed and pydot

brew install graphviz gnu-sed && \
pip install pydot

then

from networkx.drawing.nx_pydot import write_dot
write_dot(G, 'G.dot')

that will write graph like this G.dot and

gsed -e 's/label=,/label=","/g' G.dot | dot -Tpng > output.png

Hacking

pip install --editable .