统计NER中的实体标签

统计NER中的实体标签

if __name__ == '__main__':

    with open('./data/resume/train/train.txt', 'r', encoding='utf-8') as f:
        # sum = 270
        list = []
        word_ = ''
        lines = f.readlines()
        for line in lines:

            for i,words in enumerate(line):
                if words.isspace()==True or words == '\t':
                    word = line[i+1:].strip()
                    if word not in list:
                        list.append(word)

    with open('./data/resume/train/train-label.txt', 'a', encoding='utf-8') as g:
        for i in list:
            g.write(i+'\n')

    print(list)
    print(len(list))

    f.close()

数据格式如下：