CODEWITHSUNDEEP
pythondictionarynestednlpstanford-nlp
amy
amy

Reputation: 352

Tree structure from Stanford CoreNLP parser

I am trying to run StanfordCoreNLP parser and I have the following code:

from pycorenlp import StanfordCoreNLP

nlp = StanfordCoreNLP('http://localhost:9000')

def depparse(text):
    parsed=""
    output = nlp.annotate(text, properties={
      'annotators': 'depparse',
      'outputFormat': 'json'
      })

    for i in output["sentences"]:
        for j in i["basicDependencies"]:
            parsed=parsed+str(j["dep"]+'('+ j["governorGloss"]+' ')+str(j["dependentGloss"]+')'+' ')
        return parsed
text='I shot an elephant in my sleep'
depparse(text)

This gives me output as: 'ROOT(ROOT shot) nsubj(shot I) det(elephant an) dobj(shot elephant) case(sleep in) nmod:poss(sleep my) nmod(shot sleep) '

To convert the relationships into tree, I am encountered one stackoverflow post Stanford NLP parse tree format. However, the output of the parser is in "bracketed parse (tree)". Hence, I am not sure how can I achieve it. I tried changing the outputformat as well but it gives an error.

I also found Python - Generate a dictionary(tree) from a list of tuples and implemented

    list_of_tuples = [('ROOT','ROOT', 'shot'),('nsubj','shot', 'I'),('det','elephant', 'an'),('dobj','shot', 'elephant'),('case','sleep', 'in'),('nmod:poss','sleep', 'my'),('nmod','shot', 'sleep')]

nodes={}

for i in list_of_tuples:
    rel,parent,child=i
    nodes[child]={'Name':child,'Relationship':rel}

forest=[]

for i in list_of_tuples:
    rel,parent,child=i
    node=nodes[child]

    if parent=='ROOT':# this should be the Root Node
            forest.append(node)
    else:
        parent=nodes[parent]
        if not 'children' in parent:
            parent['children']=[]
        children=parent['children']
        children.append(node)

print forest

I got the following output [{'Name': 'shot', 'Relationship': 'ROOT', 'children': [{'Name': 'I', 'Relationship': 'nsubj'}, {'Name': 'elephant', 'Relationship': 'dobj', 'children': [{'Name': 'an', 'Relationship': 'det'}]}, {'Name': 'sleep', 'Relationship': 'nmod', 'children': [{'Name': 'in', 'Relationship': 'case'}, {'Name': 'my', 'Relationship': 'nmod:poss'}]}]}]

Upvotes: 0

Views: 1674

Answers (1)

Igor
Igor

Reputation: 1281

A bit off-topic indeed (this is not really an answer to your original question, but to your last comment). Posting it as an answer because the code wouldn't really fit nicely into a comment. But by just changing your depparse function slightly, you can get it in the desired format:

def depparse(text):
parsed=""
output = nlp.annotate(text, properties={
  'annotators': 'depparse',
  'outputFormat': 'json'
  })
for i in output['sentences']: # not sure if there can be multiple items here. If so, it just returns the first one currently.
    return [tuple((dep['dep'], dep['governorGloss'], dep['dependentGloss'])) for dep in i['basicDependencies']]

Upvotes: 0

Related Questions