def load_dictionary(self, source, language, dict_path):
counter, start_at = 0, int(time.time())
print("\nLoading entity_dictionary from: {}".format(dict_path))
with open(dict_path, "r", encoding="utf-8") as rf:
for line in rf:
line_arr = line.strip().split("\t\t")
if len(line_arr) != 4: continue
title, sub_title, uris, entity_id = line_arr
uris = uris.split("::;")
counter += 1
entity = Entity(entity_id, title, sub_title, source, language)
for uri in uris:
self._uri_2_id[uri] = entity_id
self._fulltitle_2_id[entity.get_full_title()] = entity_id
title_mention = self.get_mention_from_title(
entity.get_full_title())
if self._mention_2_ids.get(title_mention) is None:
self._mention_2_ids[title_mention] = dict()
self._mention_2_ids[title_mention][entity_id] = None
self.entity_dict[entity_id] = entity
print("Loaded, #{}, time: {}.".format(
counter,
str(datetime.timedelta(seconds=int(time.time()) - start_at))))
