Commit acd306f2 authored by vdberg's avatar vdberg
Browse files

sentencebysentence

parent f0163767

data/sent_df.csv

0 → 100644
+245858 −0

File added.

Preview size limit exceeded, changes collapsed.

srl.py

+19 −6
Original line number Diff line number Diff line 
from allennlp.predictors.predictor import Predictor

import pandas as pd

import logging

import logging, re





def load_nameslab():
@@ -21,6 +21,12 @@ def load_flipper(): 


namlabs_df, namlabs = load_nameslab()



def crude_ner():

    global namlabs

    nampat = '\\b' + '\\b|\\b'.join([i for i in namlabs.index]) + '\\b'

    m = re.search(nampat, sent)

    return m





def ner(t):

    global allen_ner
@@ -28,9 +34,16 @@ def ner(t): 
    return o





flipper = load_flipper()

flipper['text'] = flipper.original_title +'. '+ flipper.original_body

data = flipper[:4]

### load flipper



#flipper = load_flipper()

#flipper['text'] = flipper.original_title +'. '+ flipper.original_body



### load sent_df



sent = pd.read_csv('sent_df.csv')

sent= sent[sent.s.apply(crude_ner()) == True]

data = sent[:50]



srl_pred = Predictor.from_path("https://s3-us-west-2.amazonaws.com/allennlp/models/srl-model-2018.05.25.tar.gz")

allen_ner = Predictor.from_path("https://s3-us-west-2.amazonaws.com/allennlp/models/ner-model-2018.12.18.tar.gz")
@@ -39,10 +52,10 @@ logging.basicConfig(filename='vdberg_log.log',level=logging.DEBUG) 


verb_df = pd.DataFrame(columns=['arg', 'verb', 'same'])

for i,r in data.iterrows():

    names = ner(r.text)

    names = ner(r.s)

    logging.info('Predicted NER for {}'.format(i))



    roles = srl_pred.predict(sentence=r.text)['verbs']

    roles = srl_pred.predict(sentence=r.s)['verbs']

    logging.info('Predicted roles for {}'.format(i))

    #print('roles', roles)