我装的jieba==0.42.1可以直接对句子提取关键词，分词的部分已经封装在jieba代码里面了，修改了一下关键词提取代码： import jieba from jieba.analyse import extract_tags from jieba.analyse import textrank class KeywordModel(object): def __init__(self): jieba.load_userdict('../data/user_dict.csv') jieba.analyse.set_stop_words('../data/stopWord.txt') def get_keywords(self, sentence, type, topK=10, pos=('ns', 'n', 'vn', 'v')): """ 获取关键词 :param sentence: 文本 :param type: 使用哪种关键词算法，可选：tfidf,textrank :param topK: 获取topK关键词 :param pos: 分词保留的词性类型，eg:('ns', 'n', 'vn', 'v') :return: """ if type == 'tfidf': tfidf_keywords = extract_tags(sentence, topK=topK, allowPOS=pos) return tfidf_keywords elif type == 'textrank': textrank_keywords = textrank(sentence, topK=topK, allowPOS=pos) return textrank_keywords def keyword_interact(self, tfidf_keyword, textrank_keyword): """ 关键词交集 :param tfidf_keyword: :param textrank_keyword: :return: """ return list(set(tfidf_keyword).intersection(set(textrank_keyword))) def keyword_combine(self, tfidf_keyword, textrank_keyword): """ 关键词并集 :param tfidf_keyword: :param textrank_keyword: :param k: :return: """ combine = list(tfidf_keyword) for word in textrank_keyword: combine.append(word) return list(set(combine)) def keyword_combine_topk(self, tfidf_keyword, textrank_keyword, k): """ 关键词topk并集 :param tfidf_keyword: :param textrank_keyword: :param k: :return: """ combine = list(tfidf_keyword[:k]) for word in textrank_keyword[:k]: combine.append(word) return list(set(combine))