stance_detection.py 文件源码

def gen_training_features(self, bodies_fpath, stances_fpath):
        print 'Generating training features'
        self._train_bodies, self._train_stances = self._read(bodies_fpath, stances_fpath, True)

        print 'Generating ngrams'
        ng_start = time.time()
        self._train_unigrams = self._gen_ngrams(1, self._train_bodies, self._train_stances)
        ng_end = time.time()
        print 'ngrams generation time: ', (ng_end - ng_start), 'seconds'

        print 'Generating jaccard similarities'
        js_start = time.time()
        self.train_avg_sims, self.train_max_sims = self._gen_jaccard_sims(
                self._train_bodies,
                self._train_stances
        )
        js_end = time.time()
        print 'jaccard similarity generation time: ', (js_end - js_start), 'seconds'

        for i in range(len(self._train_stances)):
            labeled_feature = ({
                'unigrams':self._train_unigrams[i],
                'avg_sims':self.train_avg_sims[i],
                'max_sims':self.train_max_sims[i]},
                self._train_stances[i]['Stance'])
            self._labeled_feature_set.append(labeled_feature)