def test_oracle_connects_france_and_paris():
query = 'what are the major cities in france?'
target = [u'Paris']
dataset = [(query, target)]
oracle = OracleSystem(dataset)
result = oracle.execute(query)
oracle.connector.related.save_cache()
assert list(result) == target
def test_oracle_finds_chinese_language():
query = "what is the official language of china 2010?"
target = [u"Chinese language"]
dataset = [(query, target)]
oracle = OracleSystem(dataset)
result = oracle.execute(query)
oracle.connector.related.save_cache()
assert len(set(result) & set(target)) > 0
def test_timeout():
query, target = "what characters does trey parker voice?", [u"Eric Cartman", u"Fosse McDonald"]
dataset = [(query, target)]
oracle = OracleSystem(dataset)
result = oracle.execute(query)
oracle.connector.related.save_cache()
assert len(set(result) & set(target)) > 0
def test_oracle_finds_chers_son():
query, target = "what is cher's son's name?", [u"Elijah Blue Allman", u"Chaz Bono"]
dataset = [(query, target)]
oracle = OracleSystem(dataset)
result = oracle.execute(query)
oracle.connector.related.save_cache()
assert len(set(result) & set(target)) > 0
def test_oracle_connects_france_and_paris():
query = 'what are the major cities in france?'
target = [u'Paris']
dataset = [(query, target)]
oracle = OracleSystem(dataset)
result = oracle.execute(query)
oracle.connector.related.save_cache()
assert list(result) == target
def test_oracle_finds_chinese_language():
query = "what is the official language of china 2010?"
target = [u"Chinese language"]
dataset = [(query, target)]
oracle = OracleSystem(dataset)
result = oracle.execute(query)
oracle.connector.related.save_cache()
assert len(set(result) & set(target)) > 0
def test_oracle_finds_jamaican_dollar():
query = "what kind of money should i take to jamaica?"
target = [u"Jamaican dollar"]
dataset = [(query, target)]
oracle = OracleSystem(dataset)
result = oracle.execute(query)
oracle.connector.related.save_cache()
assert list(result) == target
def test_oracle_finds_ishmael():
query = "who was ishmael's mom?"
target = [u'Hagar']
dataset = [(query, target)]
oracle = OracleSystem(dataset)
result = oracle.execute(query)
oracle.connector.related.save_cache()
assert len(set(result) & set(target)) > 0
def test_oracle_finds_time_zone():
query = "what time zone am i in cleveland ohio?"
target = [u"Eastern Time Zone"]
dataset = [(query, target)]
oracle = OracleSystem(dataset)
result = oracle.execute(query)
oracle.connector.related.save_cache()
assert len(set(result) & set(target)) > 0
def test_oracle_finds_time_zone():
query = "what time zone am i in cleveland ohio?"
target = [u"Eastern Time Zone"]
dataset = [(query, target)]
oracle = OracleSystem(dataset)
result = oracle.execute(query)
oracle.connector.related.save_cache()
assert len(set(result) & set(target)) > 0
def test_oracle_finds_jamaican_dollar():
query = "what kind of money should i take to jamaica?"
target = [u"Jamaican dollar"]
dataset = [(query, target)]
oracle = OracleSystem(dataset)
result = oracle.execute(query)
oracle.connector.related.save_cache()
assert list(result) == target
def test_oracle_finds_ishmael():
query = "who was ishmael's mom?"
target = [u'Hagar']
dataset = [(query, target)]
oracle = OracleSystem(dataset)
result = oracle.execute(query)
oracle.connector.related.save_cache()
assert len(set(result) & set(target)) > 0
def test_expression():
query, target = "what character did natalie portman play in star wars?", [u"Padm\u00e9 Amidala"]
dataset = [(query, target)]
oracle = OracleSystem(dataset)
result = oracle.execute(query)
oracle.connector.related.save_cache()
print result
assert result == set(target)
def test_oracle_finds_chers_son():
query, target = "what is cher's son's name?", [
u"Elijah Blue Allman", u"Chaz Bono"
]
dataset = [(query, target)]
oracle = OracleSystem(dataset)
result = oracle.execute(query)
oracle.connector.related.save_cache()
assert len(set(result) & set(target)) > 0
def test_timeout():
query, target = "what characters does trey parker voice?", [
u"Eric Cartman", u"Fosse McDonald"
]
dataset = [(query, target)]
oracle = OracleSystem(dataset)
result = oracle.execute(query)
oracle.connector.related.save_cache()
assert len(set(result) & set(target)) > 0
def test_expression():
query, target = "what character did natalie portman play in star wars?", [
u"Padm\u00e9 Amidala"
]
dataset = [(query, target)]
oracle = OracleSystem(dataset)
result = oracle.execute(query)
oracle.connector.related.save_cache()
print result
assert result == set(target)
def get_cache_oracle_data(dataset):
oracle = OracleSystem(dataset)
i = 0
for query, target_entities in dataset:
results, expressions = oracle.get_best_results_and_expressions(query)
yield {
'query': query,
'results': results,
'expressions': expressions,
'target': target_entities,
}
i += 1
logger.info("Completed: %d", i)
if i % 10 == 0:
logger.info("Saving caches")
oracle.connector.save_cache()
logger.info("Saving complete")
def get_cache_oracle_data(dataset):
oracle = OracleSystem(dataset)
i = 0
for query, target_entities in dataset:
results, expressions = oracle.get_best_results_and_expressions(query)
yield {
'query': query,
'results': results,
'expressions': expressions,
'target': target_entities,
}
i += 1
logger.info("Completed: %d", i)
if i % 10 == 0:
logger.info("Saving caches")
oracle.connector.save_cache()
logger.info("Saving complete")
from fbsearch.dataset import get_dataset
from fbsearch import settings
from random import Random
# from log import logger
random = Random(1)
with open('nones.json') as nones_file:
dataset = [json.loads(row) for row in nones_file]
# nones = set(open('nones.txt').read().split('\n'))
# dataset_file = open(settings.DATASET_PATH)
# dataset = get_dataset(dataset_file)
# dataset = [row for row in dataset if row[0] in nones]
print dataset
# logger.info("Training")
# train_set = dataset[:2500]
# system = TensorSystem(CachedOracleSystem)
# system.train(train_set)
dataset = dataset[:100]
system = OracleSystem(dataset)
#logger.info("Testing")
# test_set = dataset[2500:]
results = get_target_and_predicted_values(dataset, system)
save(results, 'nones-output.json')
analyse('nones-output.json')