def get_question_tokens(self, index): tokens = [t.lower() for t in self._questions_data[index]['tokens']] if self._p.get("replace.entities", False): tokens = graph.replace_entities_in_instance( tokens, [self._questions_data[index]]) if self._p.get("normalize.tokens", False): tokens = [re.sub(r"\d+", "<n>", t.lower()) for t in tokens] return tokens
def get_question_tokens(self, index): tokens = [w for w, _, _ in self._dataset_tagged[index]] if self._p.get("replace.entities", False) and len(self._silver_graphs) > 0: tokens = graph.replace_entities_in_instance( tokens, [ g[0] for g in self._get_question_positive_silver(index) if len(g) > 0 ]) if self._p.get("normalize.tokens", False): tokens = [re.sub(r"\d+", "<n>", t.lower()) for t in tokens] return tokens