def tf_idf_training(comment_cnt_lower_bound, train_ratio):
# train_set, cv_set = Train.simple_partition(comment_cnt_lower_bound, \
# train_ratio)
# idf_dict = Feature.cal_idf(train_set, Config.train_idf_path)
train_set = Train.get_train_set()
Feature.cal_tf_idf(train_set, Config.train_idf_path,
Config.train_tf_idf_path, True, 200)
Genome.cal_tf_idf(Config.train_tf_idf_path, \
Config.train_genome_tf_idf_path, 200)
