def __init__(self, min_df=2, max_per=1.0, binarize=False, transform=None, replace_num='#',
source=None, subdir=None, pseudotype=None, splits_file=None, stage='training'):
name = 'property'
prefix = '_p_'
FeatureExtractorCounts.__init__(self, name=name, prefix=prefix, min_df=min_df, max_per=max_per,
binarize=binarize, transform=transform, source=source, subdir=subdir,
pseudotype=pseudotype, splits_file=splits_file, replace_num=replace_num,
stage=stage)
def __init__(self, test_fold=0, dev_subfold=None):
#print "Creating from arguments"
name = 'dataset'
prefix = '_d_'
FeatureExtractorCounts.__init__(self, name, prefix, add_oov=True,
min_doc_threshold=1,
binarize=True,
test_fold=test_fold,
dev_subfold=dev_subfold)
def __init__(self, n=1, min_df=2, max_per=1.0, binarize=False, transform=None, replace_num='#',
source=None, subdir=None, pseudotype=None, splits_file=None, stage='training',
lower=True, scale_factor=None):
name = 'ngrams'
self.n = int(n)
prefix = '_n' + str(n) + '_'
suffix = ',n=' + str(self.n)
FeatureExtractorCounts.__init__(self, name=name, prefix=prefix, min_df=min_df, max_per=max_per,
binarize=binarize, transform=transform, source=source, subdir=subdir,
pseudotype=pseudotype, splits_file=splits_file, replace_num=replace_num,
stage=stage, suffix=suffix, lower=lower, scale_factor=scale_factor)
def __init__(self, test_fold=0, dev_subfold=None, binarize=False, clusters=''):
#print "Creating from arguments"
name = 'brownclusters'
prefix = '_bc-' + clusters + '_'
FeatureExtractorCounts.__init__(self, name, prefix, add_oov=True,
min_doc_threshold=1,
binarize=binarize,
test_fold=test_fold,
dev_subfold=dev_subfold)
self.params['clusters'] = clusters
FeatureExtractorCountsBrownClusters.extend_dirname(self)
def __init__(self, test_fold=0, dev_subfold=None, n=1, min_doc_threshold=1, binarize=True,
concat_oov_counts=False, append_dataset=False, source='normalized'):
#print "Creating from arguments"
name = 'ngrams'
prefix = '_n' + str(n) + '_'
FeatureExtractorCounts.__init__(self, name, prefix, add_oov=True,
min_doc_threshold=min_doc_threshold,
binarize=binarize,
test_fold=test_fold,
dev_subfold=dev_subfold)
self.params['n'] = int(n)
self.params['concat_oov_counts'] = ast.literal_eval(str(concat_oov_counts))
self.params['append_dataset'] = ast.literal_eval(str(append_dataset))
self.params['source'] = source
FeatureExtractorCountsNgrams.extend_dirname(self)
def __init__(self, min_df=2, max_per=1.0, binarize=True, transform=None, replace_num='#',
subdir=None, source=None, splits_file=None, pseudotype=None, stage='training',
lower=True, scale_factor=None, shorten=None):
name = 'list'
prefix = '_l_'
if shorten is not None:
self.shorten = int(shorten)
if subdir == 'brown':
replace_num = None
suffix = ''
if shorten is not None:
suffix += ',shorten=' + str(self.shorten)
FeatureExtractorCounts.__init__(self, name=name, prefix=prefix, min_df=min_df, max_per=max_per,
binarize=binarize, transform=transform, replace_num=replace_num,
source=source, subdir=subdir, pseudotype=pseudotype, splits_file=splits_file,
stage=stage, lower=lower, suffix=suffix, scale_factor=scale_factor)
def from_files(cls, dirname):
#print "Loading from files", dirname
name, feature_type, test_fold, dev_subfold,\
add_oov, min_doc_threshold, binarize, extra = FeatureExtractorCounts.parse_dirname(dirname)
assert len(extra) == 1
n = int(extra[0])
assert name == 'ngrams'
return cls(test_fold=test_fold, dev_subfold=dev_subfold, n=n, min_doc_threshold=min_doc_threshold,
binarize=binarize)
