import logging
logging.basicConfig(format='%(asctime)s : %(levelname)s : %(message)s',
level=logging.INFO)
from gensim import corpora, models
texts = []
for cas in cases:
for srlSentence in cas.srlSentences:
sentence = []
for clause in srlSentence:
for role, text in clause.iteritems():
sentence.append(str((role, text)))
texts.append(sentence)
dictionary = corpora.Dictionary(texts)
corpus = [dictionary.doc2bow(text) for text in texts]
lsi = models.lsimodel.LsiModel(corpus, id2word=dictionary, num_topics=15)
if save:
lsi.save('models/lda/srlLSI')
if __name__ == "__main__":
cases = data.getAllSavedCases(senna=True)
LSIWordVec()
#srlLSI(cases, save=True)
#wordVecSrlBigrams(cases)
#cas = case.Case('6850872635911328292')
#representation = wordVecSrl(cases)
"""
for cas in cases:
print getRepresentation(cas, log=True)
"""
if 'A0' in clause and clause['A0'] == word:
polyfitInput[0] += 1
if 'V' in clause and clause['V'] == word:
polyfitInput[0] += 1
for clause in dm:
if 'A0' in clause and clause['A0'] == word:
polyfitInput[1] += 1
if 'V' in clause and clause['V'] == word:
polyfitInput[1] += 1
print polyfitInput
model[i] = np.polyfit([1, m], polyfitInput, 1)
print model
return findShortChains()
if __name__ == "__main__":
cases = data.getAllSavedCases()
labeledTraining = learning.findLabels(cases)
learning.readLabels(labeledTraining)
unlabeledCases = filter(lambda x:x not in labeledTraining, cases)
unlabeledTraining = unlabeledCases[:-1]
testing = [unlabeledCases[-1]]
learning.labelCases(labeledTraining, unlabeledTraining, testing, numIterations=20)
for cas in testing:
print 'case ' + str(cas.name)
d = {}
for person, summarySentences in cas.summary.iteritems():
personSentences = []
for sentence in summarySentences:
personSentences.append(cas.sentences[sentence])
d[person] = personSentences
for person, summarySentences in d.iteritems():
import data
from case import Case
import random
if __name__ == "__main__":
cases = data.getAllSavedCases(makeString=False, senna=False)
while True:
cas = random.choice(cases)
moreLinks = data.getAboutCases(cas)
print moreLinks
for i, link in enumerate(moreLinks):
print i
c = Case(link, senna=False)
def srlLSI(cases, log=True, save=False):
if log:
import logging
logging.basicConfig(format='%(asctime)s : %(levelname)s : %(message)s', level=logging.INFO)
from gensim import corpora, models
texts = []
for cas in cases:
for srlSentence in cas.srlSentences:
sentence = []
for clause in srlSentence:
for role, text in clause.iteritems():
sentence.append(str((role, text)))
texts.append(sentence)
dictionary = corpora.Dictionary(texts)
corpus = [dictionary.doc2bow(text) for text in texts]
lsi = models.lsimodel.LsiModel(corpus, id2word=dictionary, num_topics=15)
if save:
lsi.save('models/lda/srlLSI')
if __name__ == "__main__":
cases = data.getAllSavedCases(senna=True)
LSIWordVec()
#srlLSI(cases, save=True)
#wordVecSrlBigrams(cases)
#cas = case.Case('6850872635911328292')
#representation = wordVecSrl(cases)
"""
for cas in cases:
print getRepresentation(cas, log=True)
"""
if 'V' in clause and clause['V'] == word:
polyfitInput[0] += 1
for clause in dm:
if 'A0' in clause and clause['A0'] == word:
polyfitInput[1] += 1
if 'V' in clause and clause['V'] == word:
polyfitInput[1] += 1
print polyfitInput
model[i] = np.polyfit([1, m], polyfitInput, 1)
print model
return findShortChains()
if __name__ == "__main__":
cases = data.getAllSavedCases()
labeledTraining = learning.findLabels(cases)
learning.readLabels(labeledTraining)
unlabeledCases = filter(lambda x: x not in labeledTraining, cases)
unlabeledTraining = unlabeledCases[:-1]
testing = [unlabeledCases[-1]]
learning.labelCases(labeledTraining,
unlabeledTraining,
testing,
numIterations=20)
for cas in testing:
print 'case ' + str(cas.name)
d = {}
for person, summarySentences in cas.summary.iteritems():
personSentences = []
for sentence in summarySentences: