feature_extractor.py 文件源码

def parse_item(item):
    """

    :param item: dictionary
    :return: void
    """

    def add_school_seq(t):
        t['school_seq'] = jieba.lcut_for_search(t['school'])

    def add_title_seq(t):
        t['title_seq'] = build_tf(t['title'])[1]

    def add_abstract_seq_and_tf(t):
        t['abstract_seq_tf'], t['abstract_seq'] = build_tf(t['abstract'])

    add_abstract_seq_and_tf(item)
    add_school_seq(item)
    add_title_seq(item)

    item['_id'] = str(item['_id'])

    return item