Python sklearn.metrics 模块，jaccard_similarity_score() 实例源码

def calc(gr_truth, predicted):
    # precision, recall, fscore, _ = score(gr_truth, predicted, average='micro')
    # print('precision: {}'.format(precision))
    # print('recall: {}'.format(recall))
    # print('fscore: {}'.format(fscore))
    # print('jaccard: {}'.format(jaccard_similarity_score(gr_truth, predicted, normalize=True)))
    # print('mutual: {}'.format(mutual_info_score(gr_truth, predicted)))
    # print('mutual adj: {}'.format(adjusted_mutual_info_score(gr_truth, predicted)))
    # print('mutual norm: {}'.format(normalized_mutual_info_score(gr_truth, predicted)))
    return normalized_mutual_info_score(gr_truth, predicted)

def test_multilabel_jaccard_similarity_score():
    # Dense label indicator matrix format
    y1 = np.array([[0, 1, 1], [1, 0, 1]])
    y2 = np.array([[0, 0, 1], [1, 0, 1]])

    # size(y1 \inter y2) = [1, 2]
    # size(y1 \union y2) = [2, 2]

    assert_equal(jaccard_similarity_score(y1, y2), 0.75)
    assert_equal(jaccard_similarity_score(y1, y1), 1)
    assert_equal(jaccard_similarity_score(y2, y2), 1)
    assert_equal(jaccard_similarity_score(y2, np.logical_not(y2)), 0)
    assert_equal(jaccard_similarity_score(y1, np.logical_not(y1)), 0)
    assert_equal(jaccard_similarity_score(y1, np.zeros(y1.shape)), 0)
    assert_equal(jaccard_similarity_score(y2, np.zeros(y1.shape)), 0)

def _gen_jaccard_sims(self, bodies_dict, stances):
        # currently assumes both body and headline are longer than 0.
        punc_rem_tokenizer = nltk.RegexpTokenizer(r'\w+')

        avg_sims = []
        max_sims = []

        parsed_bodies_dict = {}
        for body_id, body in bodies_dict.iteritems():
            sents = nltk.sent_tokenize(body)
            sents = self._remove_punctuation(sents)
            sents = self._word_tokenize(sents)
            parsed_bodies_dict[body_id] = sents # cache parsed body


        for st in stances:
            headline = st['Headline']
            headline = headline.translate(self.REMOVE_PUNC_MAP)
            headline = nltk.word_tokenize(headline)

            jacc_sims = []
            for sent in sents:
                if len(sent) < 1:
                    continue
                # extend shorter word list so that both are the same length
                len_diff = len(headline) - len(sent)
                headline_cpy = headline
                sent_cpy = sent

                if len_diff < 0: # sent longer than headline
                    headline_cpy = headline_cpy + ([headline_cpy[-1]] * abs(len_diff))
                elif len_diff > 0: # headline longer than sent
                    sent_cpy = sent_cpy + ([sent_cpy[-1]] * abs(len_diff))

                jacc_sims.append(jaccard_similarity_score(headline_cpy, sent_cpy))

            avg_sim = self._threshold_parser((sum(jacc_sims) / len(jacc_sims)), [0.2])
            max_sim = self._threshold_parser(max(jacc_sims), [0.2])
            avg_sims.append(avg_sim)
            max_sims.append(max_sim)

        return avg_sims, max_sims