from conceptnet5.edges import make_edge
from conceptnet5.whereami import get_project_filename
FRAME_DATA = json.load(
codecs.open(get_project_filename('data/info/zh_frames.json'))
)
def handle_raw_assertion(line):
if not line:
return
parts = line.split(', ')
user, frame_id, concept1, concept2 = parts
fdata = FRAME_DATA[frame_id]
ftext = fdata['text']
rel = fdata['relation']
surfaceText = ftext.replace(u'{1}', u'[['+concept1+u']]').replace(u'{2}', u'[['+concept2+u']]')
start = make_concept_uri(concept1, 'zh_TW')
end = make_concept_uri(concept2, 'zh_TW')
sources = ['/s/activity/ptt/petgame', '/s/contributor/petgame/' + user]
edge = make_edge(rel, start, end, dataset='/d/conceptnet/4/zh',
license='/l/CC/By', sources=sources,
surfaceText=surfaceText, weight=1)
yield json.dumps(edge, ensure_ascii=False)
if __name__ == '__main__':
from conceptnet5.readers import transform_stream
transform_stream(handle_raw_assertion)
contributors = [
s for s in source_list
if s.startswith('/s/contributor')
]
assert len(contributors) <= 1, contributors
edge = make_edge(relation,
start,
end,
dataset,
LICENSE,
source_list,
'/ctx/all',
frame_text,
weight=weight)
okay = True
if contributors:
uri = edge['uri']
contributor = contributors[0]
if (uri, contributor) in self.seen_sources:
okay = False
else:
self.seen_sources.add((uri, contributor))
if okay:
yield json.dumps(edge, ensure_ascii=False)
if __name__ == '__main__':
from conceptnet5.readers import transform_stream
builder = CN4Builder()
transform_stream(builder.handle_raw_assertion)
def handle_raw_assertion(line):
if not line:
return
parts = line.split(', ')
user, frame_id, concept1, concept2 = parts
fdata = FRAME_DATA[frame_id]
ftext = fdata['text']
rel = fdata['relation']
surfaceText = ftext.replace(u'{1}', u'[[' + concept1 + u']]').replace(
u'{2}', u'[[' + concept2 + u']]')
start = make_concept_uri(concept1, 'zh_TW')
end = make_concept_uri(concept2, 'zh_TW')
sources = ['/s/activity/ptt/petgame', '/s/contributor/petgame/' + user]
edge = make_edge(rel,
start,
end,
dataset='/d/conceptnet/4/zh',
license='/l/CC/By',
sources=sources,
surfaceText=surfaceText,
weight=1)
yield json.dumps(edge, ensure_ascii=False)
if __name__ == '__main__':
from conceptnet5.readers import transform_stream
transform_stream(handle_raw_assertion)
dataset = build_data_set(parts_dict)
sources = build_sources(parts_dict, preposition_fix)
reject = False
for source_list, weight in sources:
if 'commons2_reject' in ' '.join(source_list):
reject = True
if not reject:
for source_list, weight in sources:
if not by_bedume_and_bad(source_list,start,end):
contributors = [s for s in source_list if s.startswith('/s/contributor')]
assert len(contributors) <= 1, contributors
edge = make_edge(relation, start, end, dataset, LICENSE, source_list, '/ctx/all', frame_text, weight=weight)
okay = True
if contributors:
uri = edge['uri']
contributor = contributors[0]
if (uri, contributor) in self.seen_sources:
okay = False
else:
self.seen_sources.add((uri, contributor))
if okay:
yield json.dumps(edge, ensure_ascii=False)
if __name__ == '__main__':
from conceptnet5.readers import transform_stream
builder = CN4Builder()
transform_stream(builder.handle_raw_assertion)
