# fix random seed to ensure reproducibility
random.seed(42)
# parse command line arguments
parser = argparse.ArgumentParser(
description='rule extraction by label counting')
parser.add_argument('-q',
'--quiet',
action="store_true",
help='disables info output')
parser.add_argument('config_file', help='the config file to use')
parser.add_argument('config_name', help='the name of the configuration')
args = parser.parse_args()
# read config file
config = util.parse_config_file(args.config_file, args.config_name)
# dictionary for the confidence values extracted from the data set
# maps from rule type to a list containing rules and their confidence
rules = {}
for rule_type in util.rule_types:
rules[rule_type] = []
if not args.quiet:
print("Looking for rules involving two concepts...")
# first look for simple rules involving only two concepts
# like "first_concept IMPLIES second_concept" and "first_concept IS DIFFERENT FROM second_concept"
for first_concept in config["concepts"]:
if not args.quiet:
print(first_concept)
# -*- coding: utf-8 -*- """ Analyzes the given data set and prints out its characteristics. 1st argument: config file, 2nd argument: config_name Created on Thu Feb 8 09:33:14 2018 @author: lbechberger """ import util, sys config = util.parse_config_file(sys.argv[1], sys.argv[2]) util.data_set_characteristics(config["training_vectors"], config["validation_vectors"], config["test_vectors"], config["concepts"])