def test_ner_quality(self):
print "Evaluating NERs quality against CoNLL 2002 baseline"
c = CoNLL2002.testb("nl")
b = c.baseline()
self.eval_ner("lingpipe", c, b)
self.eval_ner("stanford", c, b)
import platform
if platform.platform().lower().find("linux") >= 0:
self.eval_ner("tnt", c, b)
else:
print "Skip evaluating TNT tagger. It is possible on Linux platform only."
def test_ner_quality(self):
print "Evaluating NERs quality against CoNLL 2002 baseline"
c = CoNLL2002.testb("nl")
b = c.baseline()
self.eval_ner("lingpipe", c, b)
self.eval_ner("stanford", c, b)
import platform
if platform.platform().lower().find("linux") >= 0:
self.eval_ner("tnt", c, b)
else:
print "Skip evaluating TNT tagger. It is possible on Linux platform only."
def test_conll2002(self):
def f(): CoNLL2002.testa('d')
self.assertRaises(BaseException, f)
def f(): CoNLL2002.testb('d')
self.assertRaises(BaseException, f)
def f(): CoNLL2002.train('d')
self.assertRaises(BaseException, f)
c = CoNLL2002.testa("nl")
b = c.baseline()
self.assertEqual(b['phrases'], len([ e for e in c.ne_tokens()]))
self._test_corpuses(CoNLL2002, "nl")
self._test_corpuses(CoNLL2002, "es")
def test_conll2002(self):
def f():
CoNLL2002.testa('d')
self.assertRaises(BaseException, f)
def f():
CoNLL2002.testb('d')
self.assertRaises(BaseException, f)
def f():
CoNLL2002.train('d')
self.assertRaises(BaseException, f)
c = CoNLL2002.testa("nl")
b = c.baseline()
self.assertEqual(b['phrases'], len([e for e in c.ne_tokens()]))
self._test_corpuses(CoNLL2002, "nl")
self._test_corpuses(CoNLL2002, "es")
def test_eval(self):
c = CoNLL2000.testa("en")
t = [e for e in c.iob_tokens('SYN')]
all_t = [e for e in c.tokens()]
r = c.conlleval(t)
self._test_self_eval_result(r, all_t, t)
self._test_self_eval(CoNLL2002.testa("nl"))
self._test_self_eval(CoNLL2002.testb("nl"))
self._test_self_eval(CoNLL2002.train("nl"))
self._test_self_eval(CoNLL2002.testa("es"))
self._test_self_eval(CoNLL2002.testb("es"))
self._test_self_eval(CoNLL2002.train("es"))
self._test_self_eval(CoNLL2003.testa("en"))
self._test_self_eval(CoNLL2003.testb("en"))
self._test_self_eval(CoNLL2003.train("en"))
def test_eval(self):
c = CoNLL2000.testa("en")
t = [ e for e in c.iob_tokens('SYN') ]
all_t = [e for e in c.tokens() ]
r = c.conlleval(t)
self._test_self_eval_result(r, all_t, t)
self._test_self_eval(CoNLL2002.testa("nl"))
self._test_self_eval(CoNLL2002.testb("nl"))
self._test_self_eval(CoNLL2002.train("nl"))
self._test_self_eval(CoNLL2002.testa("es"))
self._test_self_eval(CoNLL2002.testb("es"))
self._test_self_eval(CoNLL2002.train("es"))
self._test_self_eval(CoNLL2003.testa("en"))
self._test_self_eval(CoNLL2003.testb("en"))
self._test_self_eval(CoNLL2003.train("en"))
t2 = load_all_recognized_tokens('lingpipe', corpus.lang, model)
t3 = load_all_recognized_tokens('stanford', corpus.lang, model)
r = intersect_tokens_arrays(t1, 0.25, t2, 0.25, t3, 0.5)
stat = EntityIntersectionStat(r,[ e for e in corpus.ne_tokens() ], "parametrized votes")
print stat
return r
from gravity.tae.corpora.conll import CoNLL2002
from gravity.tae.corpora.conll import CoNLL2003
#corpus = CoNLL2003.testb("en")
corpus = CoNLL2002.testb("nl")
#show_corpus_characterestics(corpus)
#corpus = CoNLL2003.testb("en")
#corpus = CoNLL2002.testb("es")
# r = calc_parametrized_vote_result(corpus)
# print len(r)
evaluate('stanford2', corpus)
#save_nose_entities('stanford', corpus)
def f():
CoNLL2002.train('d')
def f():
CoNLL2002.testb('d')
def f():
CoNLL2002.testa('d')
def f():
CoNLL2002.path('ddd')
def f(): CoNLL2002.train('d')
self.assertRaises(BaseException, f)
def f(): CoNLL2002.path('ddd')
self.assertRaises(BaseException, f)
