def test6(ntrees=10, steps=10):
import wsj10
tb = wsj10.WSJ10()
tb.trees = tb.trees[:ntrees]
m = DMV(tb)
m.train(steps)
return m
def test0(ntrees=10, steps=1, tb=None, clazz=None):
if tb is None:
import wsj10
tb = wsj10.WSJ10()
if clazz is None:
clazz = DMVCCM
tb.trees = tb.trees[:ntrees]
m = clazz(tb)
t0 = time.clock()
m.train(steps)
t = time.clock() - t0
print('Tiempo (seg.):', t)
return m
def main1():
import wsj10
tb = wsj10.WSJ10()
m = UBound(tb)
m.eval()
return m
def _get_treebank(self, treebank=None):
if treebank is None:
treebank = wsj10.WSJ10()
return treebank
# print m
# m.train(2)
# m.train(10)
# m.test()
# can't do - presumably not tagged
#m = ccm.CCM(dependency_treebank)
#from nltk.corpus import treebank
#t = treebank.parsed_sents('wsj_0001.mrg')[0]
tb = wsj10.WSJ10(basedir=CORPUS_DIR)
#print tb.get_trees()
#print tb.raw()
tb.print_stats()
m = ccm.CCM(tb)
print m
m.train(40)
m.test()
# with 10 iterations
# Sentences: 7422
# Micro-averaged measures:
# Precision: 60.5
# Recall: 76.8
# Harmonic mean F1: 67.6
def main1():
import wsj10
tb = wsj10.WSJ10()
m = LBranch(tb)
m.eval()