BM25.py 文件源码

def bm25(p, titles, answers, scores):
    original_titles = copy.deepcopy(titles)
    titles = [remove_punctuation_re(title) for title in titles]
    answers = [remove_punctuation_re(answer) for answer in answers]
    p = remove_punctuation_re(p)
    titles = [' '.join(jieba.cut(title)) for title in titles]
    p = ' '.join(jieba.cut(p))
    wordindoc, wordindata, doclen, sumlen = init(titles, False)
    global avglen
    avglen = 1.0 * sumlen / N
    res = search(p, zip(titles, original_titles, answers, scores), wordindoc, wordindata, doclen)
    titles, answers, scores = [], [], []
    for key, _ in res:
        titles.append(key[0])
        answers.append(key[1])
        scores.append(key[2])
    return titles, answers, scores