def __init__(self):
self.nlp = spacy.load('en_core_web_sm')
self.db = MultiWozDB(cfg.dbs)
data_path = 'data/multi-woz/annotated_user_da_with_span_full.json'
archive = zipfile.ZipFile(data_path + '.zip', 'r')
self.convlab_data = json.loads(archive.open(data_path.split('/')[-1], 'r').read().lower())
self.delex_sg_valdict_path = 'data/multi-woz-processed/delex_single_valdict.json'
self.delex_mt_valdict_path = 'data/multi-woz-processed/delex_multi_valdict.json'
self.ambiguous_val_path = 'data/multi-woz-processed/ambiguous_values.json'
self.delex_refs_path = 'data/multi-woz-processed/reference_no.json'
self.delex_refs = json.loads(open(self.delex_refs_path, 'r').read())
if not os.path.exists(self.delex_sg_valdict_path):
self.delex_sg_valdict, self.delex_mt_valdict, self.ambiguous_vals = self.get_delex_valdict()
else:
self.delex_sg_valdict = json.loads(open(self.delex_sg_valdict_path, 'r').read())
self.delex_mt_valdict = json.loads(open(self.delex_mt_valdict_path, 'r').read())
self.ambiguous_vals = json.loads(open(self.ambiguous_val_path, 'r').read())
self.vocab = utils.Vocab(cfg.vocab_size)
self.slot_list = [
'hotel-pricerange', 'hotel-type', 'hotel-parking', 'hotel-stay', 'hotel-day', 'hotel-people', \
'hotel-area', 'hotel-stars', 'hotel-internet', 'train-destination', 'train-day', 'train-departure', 'train-arrive', \
'train-people', 'train-leave', 'attraction-area', 'restaurant-food', 'restaurant-pricerange', 'restaurant-area', \
'attraction-name', 'restaurant-name', 'attraction-type', 'hotel-name', 'taxi-leave', 'taxi-destination', 'taxi-departure', \
'restaurant-time', 'restaurant-day', 'restaurant-people', 'taxi-arrive', "hospital-department"
]
self.gating_dict = {'ptr': 0, 'dontcare': 1, 'none': 2}
def __init__(self):
self.nlp = spacy.load('en_core_web_sm')
self.db = MultiWozDB(cfg.dbs)
data_path = 'data/multi-woz/annotated_user_da_with_span_full.json'
archive = zipfile.ZipFile(data_path + '.zip', 'r')
self.convlab_data = json.loads(
archive.open(data_path.split('/')[-1],
'r').read().decode('utf-8').lower())
self.delex_sg_valdict_path = 'data/multi-woz-processed/delex_single_valdict.json'
self.delex_mt_valdict_path = 'data/multi-woz-processed/delex_multi_valdict.json'
self.ambiguous_val_path = 'data/multi-woz-processed/ambiguous_values.json'
self.delex_refs_path = 'data/multi-woz-processed/reference_no.json'
self.delex_refs = json.loads(open(self.delex_refs_path, 'r').read())
if not os.path.exists(self.delex_sg_valdict_path):
self.delex_sg_valdict, self.delex_mt_valdict, self.ambiguous_vals = self.get_delex_valdict(
)
else:
self.delex_sg_valdict = json.loads(
open(self.delex_sg_valdict_path, 'r').read())
self.delex_mt_valdict = json.loads(
open(self.delex_mt_valdict_path, 'r').read())
self.ambiguous_vals = json.loads(
open(self.ambiguous_val_path, 'r').read())
self.vocab = utils.Vocab(cfg.vocab_size)
def __init__(self, vocab=None):
super().__init__()
self.nlp = spacy.load('en_core_web_sm')
self.db = MultiWozDB(cfg.dbs)
self.domain_files = json.loads(open(cfg.domain_file_path, 'r').read())
self.slot_value_set = json.loads(
open(cfg.slot_value_set_path, 'r').read())
if cfg.multi_acts_training:
self.multi_acts = json.loads(open(cfg.multi_acts_path, 'r').read())
test_list = [
l.strip().lower() for l in open(cfg.test_list, 'r').readlines()
]
dev_list = [
l.strip().lower() for l in open(cfg.dev_list, 'r').readlines()
]
self.dev_files, self.test_files = {}, {}
for fn in test_list:
self.test_files[fn.replace('.json', '')] = 1
for fn in dev_list:
self.dev_files[fn.replace('.json', '')] = 1
self.exp_files = {}
if 'all' not in cfg.exp_domains:
for domain in cfg.exp_domains:
fn_list = self.domain_files.get(domain)
if not fn_list:
raise ValueError('[%s] is an invalid experiment setting' %
domain)
for fn in fn_list:
self.exp_files[fn.replace('.json', '')] = 1
if vocab:
self.vocab = vocab
self.vocab_size = vocab.size
else:
self.vocab_size = self._build_vocab()
self._load_data()
if cfg.limit_bspn_vocab:
self.bspn_masks = self._construct_bspn_constraint()
if cfg.limit_aspn_vocab:
self.aspn_masks = self._construct_aspn_constraint()
self.multi_acts_record = None