semantic.py 文件源码

def synonym_cut(sentence, pattern="wf"):
    """Cut the sentence into a synonym vector tag.
    ??????????????

    If a word in this sentence was not found in the synonym dictionary,
    it will be marked with default value of the word segmentation tool.
    ????????????????????????

    Args:
        pattern: 'w'-??, 'k'-??????'t'-?????, 'wf'-????, 'tf-?????'?
    """
    sentence = sentence.rstrip(tone_words)
    synonym_vector = []
    if pattern == "w":
        result = list(jieba.cut(sentence))
        synonym_vector = [item for item in result if item not in punctuation_all]
    elif pattern == "k":
        synonym_vector = analyse.extract_tags(sentence, topK=1)
    elif pattern == "t":
        synonym_vector = analyse.extract_tags(sentence, topK=10)
    elif pattern == "wf":
        result = posseg.cut(sentence)
        # synonym_vector = [(item.word, item.flag) for item in result \
        # if item.word not in punctuation_all]
        # Modify in 2017.4.27 
        for item in result:
            if item.word not in punctuation_all:
                if len(item.flag) < 4:
                    item.flag = list(posseg.cut(item.word))[0].flag
                synonym_vector.append((item.word, item.flag))
    elif pattern == "tf":
        result = posseg.cut(sentence)
        tags = analyse.extract_tags(sentence, topK=10)
        for item in result:
            if item.word in tags:
                synonym_vector.append((item.word, item.flag))
    return synonym_vector