def __init__(self):
"""
Default constructor that calls evaluate_wmd_model() from within
and prints out model accuracy when call to evaluate_wmd_model() is returned.
"""
# parse MSR data
test_data, sent_dict, pair_dict = parse()
# word mover model -- take long to load the model!
wm_model = WMD()
# copnvert the ID->String dict to ID-> token dict
candidate_dict = wmd_utils.sent_dict_to_tok_dict(sent_dict)
wm_model.evaluate_model(candidate_dict, pair_dict)
def __init__(self):
'''
Create db connection and
load data from the local folder.
'''
self.client = MongoClient('db.infinity.buda.link', 27017,
connect=False)
self.db = self.client.styria
self.documents = self.db.documents
count = self.documents.count()
if count <= 0:
files = parser.parse('../20news-18828', 'iso-8859-1')
for file_name, file_content in files.items():
self.insert(file_content)
def __init__(self):
'''
Create db connection and
load data from the local folder.
'''
self.client = MongoClient('db.infinity.buda.link',
27017,
connect=False)
self.db = self.client.styria
self.documents = self.db.documents
count = self.documents.count()
if count <= 0:
files = parser.parse('../20news-18828', 'iso-8859-1')
for file_name, file_content in files.items():
self.insert(file_content)
if __name__ == '__main__':
# defaults
QUERY = 'test case'
ALGORITHM = 'bag_of_words'
# ALGORITHM = 'vector_space'
# ALGORITHM = 'binary_independence'
RESULTS = 20
DOCUMENT = ''
# load local files
# files_path = 'test-small/subset'
# files_path = '20news-18828/alt.atheism'
files_path = '20news-18828'
logger.info('Loading files...')
files = parse(files_path, 'iso-8859-1')
logger.info('Files from %s are loaded.' % files_path)
# create algorithm box object (context object)
algorithm_box = AlgorithmBox()
algorithm_box.files = files
print()
print('---- MANUAL ----------')
print('-q "query" DEFAULT: "test case"')
print('-a "algorithm" -> bag_of_words, vector_space, binary_independence')
print(' DEFAULT: bag_of_words')
print('-n "number of results" DEFAULT: 20')
print('-d "new document" DEFAULT: ""')
print('-e exit')
print('----------------------')
if __name__ == '__main__':
# defaults
QUERY = 'test case'
ALGORITHM = 'bag_of_words'
# ALGORITHM = 'vector_space'
# ALGORITHM = 'binary_independence'
RESULTS = 20
DOCUMENT = ''
# load local files
# files_path = 'test-small/subset'
# files_path = '20news-18828/alt.atheism'
files_path = '20news-18828'
logger.info('Loading files...')
files = parse(files_path, 'iso-8859-1')
logger.info('Files from %s are loaded.' % files_path)
# create algorithm box object (context object)
algorithm_box = AlgorithmBox()
algorithm_box.files = files
print()
print('---- MANUAL ----------')
print('-q "query" DEFAULT: "test case"')
print('-a "algorithm" -> bag_of_words, vector_space, binary_independence')
print(' DEFAULT: bag_of_words')
print('-n "number of results" DEFAULT: 20')
print('-d "new document" DEFAULT: ""')
print('-e exit')
print('----------------------')
