def test_cluster1():
""" Testing for cluster, using test data
"""
cluster = Cluster(epoch=32, size=500, tokenizer="tokenize")
pipeline = Pipeline()
for item in pipeline.capture_item():
cluster.put_item(item)
cluster.cluster()
# assert '캠프' in list(map(list, zip(*cluster.similar('노무현'))))[0]
# assert '사건' in list(map(list, zip(*cluster.similar('박근혜'))))[0]
assert len(cluster.clusters) == len(test_text)
assert cluster.vectors.shape == (len(test_text), 500)
assert len(cluster.unique) <= len(test_text)
assert len(cluster.unique) == len(cluster.dumps)
for dump in cluster.dumps:
items, vectors, counter = zip(*dump)
for item in items:
assert isinstance(item, Item)
pipeline.dress_item(items)
def test_app3():
""" Testing for cluster, using test data
"""
cluster = Cluster(tokenizer="tokenize")
pipe = PipelineCsv(test_csv)
for item in pipe.capture_item():
cluster.put_item(item)
cluster.cluster()
extractor = Extractor(cluster)
for idx, dump in enumerate(cluster.dumps):
items, vectors, counter = map(list, zip(*dump))
extracted = extractor.dump(idx)
pipe.dress_item(extracted)
print (cluster.distribution)
def test_app1():
""" Testing for cluster, using test data
"""
cluster = Cluster(epoch=32, tokenizer="stemize")
pipeline = PipelineFile()
for item in pipeline.capture_item():
cluster.put_item(item)
cluster.cluster()
extractor = Extractor(cluster)
for idx, dump in enumerate(cluster.dumps):
items, vectors, counter = map(list, zip(*dump))
extracted = extractor.dump(idx)
assert isinstance(extracted.keywords, list)
pipeline.dress_item(extracted)
def test_extractor1():
cluster = Cluster(epoch=32, tokenizer="tokenize")
pipeline = Pipeline()
for item in pipeline.capture_item():
cluster.put_item(item)
cluster.cluster()
extractor = Extractor(cluster)
for idx, dump in enumerate(cluster.dumps):
items, vectors, counter = map(list, zip(*dump))
assert set(['items', 'vectors', 'counter', 'center',
'keywords']) == set(extractable.s.keys())
extracted = extractor.dump(idx)
assert isinstance(extracted, Item)
assert isinstance(extracted.keywords, list)
assert 32 == len(extracted.keywords)