portion = 0
if len(sys.argv) > 1:
portion = int(sys.argv[1])
fg = FeatureGenerator()
Xs, ys = fg.generate_training("emnlp")
classifier = LogisticRegressionClassifier()
classifier.train(Xs, ys)
import csv
base = "data/emnlp"
rels = []
tables = load_collection("emnlp")
from collections import defaultdict
import re
def fact_check_and_test(q, rel):
question = Question(text=q, type="NUM")
tuples, q_features = fg.generate_test(tables, question)
q_match = False
matches = dict()
p_match = 0.0
found_match = False
total_geq = 0
s = pickle.dumps(lr)
print(lr)
print("Trained")
queries = [
"Hamas was founded in 1985",
"In America, in June 1901, the average temperature was 16.6C",
"The World life expectancy was 52 in 1960",
"world life expectancy rose sharply to 80 in 2014",
"In 2012 there were 3,282,570 bee colonies in America",
"In 2016, the USA contributed $550bn to the financial intermediary funds"
]
tables = load_collection("herox")
for question in queries:
q = Question(text=question, type="NUM")
q.parse()
q_match = False
tuples = []
for obj in q.nps.union(q.nes):
tuples.extend(get_all_tuples(tables, obj))
done_tuple = False
for tuple in tuples:
done_tuple = True
table_name = tuple[0]
from tabular.filtering import load_collection
from tabular.tuples import get_all_tuples
if __name__ == "__main__":
tables = load_collection("training")
print(get_all_tuples(tables, "Japan"))