def __init__(self, basedir, svmdir, type="dev"):
self.basedir = basedir
self.svmdir = svmdir
self.type = type
self.svm_args = "-t 5"
self.training_data = get_original_data("resources/cleaned-train.gold")
self.test_data = get_original_data("resources/cleaned-{:s}.notag".format(type))
parser.add_argument("-a",
"--answers",
help="the input file has the answers",
action="store_true")
parser.add_argument("-m",
"--mallet",
help="prepare mallet output",
action="store_true")
all_args = parser.parse_args()
tree_funcs = feature_list_reader(all_args.tree_list, locals())
feature_funcs = feature_list_reader(all_args.feature_list, locals())
if all_args.mallet:
feature_funcs.insert(0, relation_type)
data = get_original_data(all_args.input_file)
f = Featurizer(data, tree_funcs, feature_funcs, not all_args.answers)
if all_args.mallet:
f.build_mallet_features()
f.write_no_tag(all_args.output_dir, all_args.file_suffix)
else:
f.build_features()
if all_args.answers:
f.build_relation_class_vectors()
f.write_multiple_vectors(all_args.output_dir, all_args.file_suffix)
else:
f.write_no_tag(all_args.output_dir, all_args.file_suffix)
print "built your new feature vectors at {}".format(
all_args.output_dir)
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument("input_file")
parser.add_argument("output_dir")
parser.add_argument("file_suffix")
parser.add_argument("tree_list")
parser.add_argument("feature_list")
parser.add_argument("-a", "--answers", help="the input file has the answers", action="store_true")
parser.add_argument("-m", "--mallet", help="prepare mallet output", action="store_true")
all_args = parser.parse_args()
tree_funcs = feature_list_reader(all_args.tree_list, locals())
feature_funcs = feature_list_reader(all_args.feature_list, locals())
if all_args.mallet:
feature_funcs.insert(0, relation_type)
data = get_original_data(all_args.input_file)
f = Featurizer(data, tree_funcs, feature_funcs, not all_args.answers)
if all_args.mallet:
f.build_mallet_features()
f.write_no_tag(all_args.output_dir, all_args.file_suffix)
else:
f.build_features()
if all_args.answers:
f.build_relation_class_vectors()
f.write_multiple_vectors(all_args.output_dir, all_args.file_suffix)
else:
f.write_no_tag(all_args.output_dir, all_args.file_suffix)
print "built your new feature vectors at {}".format(all_args.output_dir)
