model.py 文件源码

def load_data(filenames):
    """ ???????????
    :param filenames: ??????
    :return: Bunch ????. See:
        http://scikit-learn.org/stable/datasets/index.html#datasets
    """
    # ??????
    data = []
    # ???????????
    target = []
    # ????
    target_names = {}
    # ????????????HUM,???????
    data_re = re.compile(r'(\w+),(.+)')

    for line in fileinput.input(filenames):
        match = data_re.match(line.decode('utf-8'))
        if not match:
            raise Exception("Invalid format in dataset {} at line {}"
                            .format(fileinput.filename(),
                                    fileinput.filelineno()))

        label, text = match.group(1), match.group(2)

        if label not in target_names:
            target_names[label] = len(target_names)
        # ??????????`HUM`, `LOC`, etc.?
        target.append(label)
        # ?????????????{'HUM': 1, 'LOC': 2}?
        # target.append(target_names[label])
        data.append(text)

    return Bunch(
        data=numpy.array(data),
        target=numpy.array(target),
        target_names=numpy.array([k for k in target_names]),
    )