def simple_tagger_eval(eval_path, gold_path, out_f = sys.stdout, csv=True):
eval_c = POSCorpus.read_simpletagger(eval_path)
gold_c = POSCorpus.read_simpletagger(gold_path)
poseval(eval_c, gold_c, out_f)
def conll_to_slashtags(infiles, outpath):
'''
This will
@param infiles: A list of CONLL pathnames to convert.
@param outpath: A single output pathname for the slashtags.
'''
main_c = POSCorpus()
for f in infiles:
from intent.corpora.conll import ConllCorpus
cp = ConllCorpus.read(f)
main_c.extend(c)
st = c.slashtags('/', lowercase=True)
# Create the containing path if it doesn't already exist
os.makedirs(os.path.dirname(outpath), exist_ok=True)
of = open(outpath, 'w', encoding='utf-8')
of.write(st)
of.close()
def slashtags_eval(goldpath, testpath, delimiter, out_f=sys.stdout, tagmap=None, matrix=False, details=False, length_limit=None):
'''
Evaluate a "slashtags" format file
:param goldpath:
:type goldpath:
:param testpath:
:type testpath:
:param delimiter:
:type delimiter:
:param out_f:
:type out_f:
:param tagmap:
:type tagmap:
:rtype: POSEvalDict
'''
gold_c = POSCorpus.read_slashtags(goldpath)
test_c = POSCorpus.read_slashtags(testpath)
poseval(test_c, gold_c, out_f, matrix=matrix, details=details, length_limit=length_limit)
def slashtags_to_simpletagger(in_path, out_path):
p = POSCorpus.read_slashtags(in_path)
out_f = open(out_path, 'w', encoding='utf-8')
for inst in p:
sf = SequenceFeature(inst)
while sf:
out_f.write('%s ' % sf.form)
# out_f.write('word-%s ' % sf.form)
# out_f.write('pre-3-%s ' % sf.prefix(3))
# out_f.write('pre-2-%s ' % sf.prefix(2))
#
# out_f.write('suf-3-%s ' % sf.suffix(3))
# out_f.write('suf-2-%s ' % sf.suffix(2))
#
# #===================================================================
# # Context Features
# #===================================================================
# out_f.write('prev-%s ' % sf.prev().form)
# out_f.write('next-%s ' % sf.next().form)
#
# #===================================================================
# # More Context
# #===================================================================
# out_f.write('prev-prev-%s ' % sf.prev().prev().form)
# out_f.write('next-next-%s ' % sf.next().next().form)
# Finally, write out the label
out_f.write('%s\n' % sf.label)
sf = sf.next()
out_f.write('\n')
out_f.close()
