test.py 文件源码

def run(self):
        for d_type, datasets in self.sim_datasets.iteritems():
            for data, fn in datasets:
                logging.info(
                    'testing on data {0} of type {1} ({2} pairs)'.format(
                        fn, d_type, len(data.pairs)))
                for e_type, models in self.e_models.iteritems():
                    for model, fn in models:
                        logging.info(
                            '\ttesting embedding {0} of type {1}'.format(
                                fn, e_type))
                        answers, gold_sims, oovs = [], [], 0
                        for (w1, w2), gold in data.pairs.iteritems():
                            sim = model.get_sim(w1, w2)
                            if sim:
                                answers.append(sim)
                                gold_sims.append(gold)
                            else:
                                oovs += 1
                        corr = spearmanr(answers, gold_sims)
                        logging.info('Spearman correlation: {0}'.format(corr))
                        logging.info('pairs skipped (OOVs): {0}'.format(oovs))