def features_gen(config):
"""
Reads the training data set and yields feature sets associated with class labels
"""
ratingsmean = get_ratingsmean(config)
logging.debug("middle rating value used for training = %d"%ratingsmean)
sourcetype = config['training']['type']
path = join( config['data_path'], config['training']['path'] )
dataclass = _dynamic_get_class("mpdata", sourcetype)
data = dataclass( path, dialect="excel")
for com in data:
if 'rating' in com:
if com['rating'] >= ratingsmean:
yield (word_feats(mptokenize.tokenize(com['body'])), 'pos')
else:
yield (word_feats(mptokenize.tokenize(com['body'])), 'neg')
logging.debug("total of bad lines : %d"%data.impossible_line)
def features_gen(config):
"""
Reads the training data set and yields feature sets associated with class labels
"""
ratingsmean = get_ratingsmean(config)
logging.debug("middle rating value used for training = %d" % ratingsmean)
sourcetype = config['training']['type']
path = join(config['data_path'], config['training']['path'])
dataclass = _dynamic_get_class("mpdata", sourcetype)
data = dataclass(path, dialect="excel")
for com in data:
if 'rating' in com:
if com['rating'] >= ratingsmean:
yield (word_feats(mptokenize.tokenize(com['body'])), 'pos')
else:
yield (word_feats(mptokenize.tokenize(com['body'])), 'neg')
logging.debug("total of bad lines : %d" % data.impossible_line)
def articles_gen(config, dbconnect):
content_gen = dbconnect.get_contents()
total=0
try:
while 1:
alltokens = []
(id, contents) = content_gen.next()
# tokenizes all selected fields from DB
for field in contents:
alltokens += mptokenize.tokenize( nltk.clean_html(field) )
yield (id, alltokens)
total+=1
except StopIteration:
logging.debug( "analyzed %d articles"%total )
def articles_gen(config, dbconnect):
content_gen = dbconnect.get_contents()
total = 0
try:
while 1:
alltokens = []
(id, contents) = content_gen.next()
# tokenizes all selected fields from DB
for field in contents:
alltokens += mptokenize.tokenize(nltk.clean_html(field))
yield (id, alltokens)
total += 1
except StopIteration:
logging.debug("analyzed %d articles" % total)
