Datasets:
Tasks:
Translation
Multilinguality:
translation
Size Categories:
1K<n<10K
Language Creators:
crowdsourced
Annotations Creators:
expert-generated
Source Datasets:
original
Tags:
License:
File size: 9,943 Bytes
3d81c4a 8298e0a 3d81c4a 2d13342 0abbaee 2d13342 3d81c4a b05395d 3d81c4a c387054 3d81c4a b05395d 3d4cc61 8236ab8 3d4cc61 3d81c4a b05395d 3d81c4a 3d4cc61 2608406 3ad179d 2608406 c0173f7 2608406 c0173f7 ba896b7 c0173f7 5cbedcb c0173f7 ba896b7 c0173f7 2608406 5157745 2608406 3ad179d 6821ffd 2608406 5157745 a6a5789 3d81c4a 97fadc5 3d81c4a 2d13342 3d81c4a 2666d83 3d81c4a 90f3403 4f142b4 b2b5ad7 90f3403 0278bca 0140790 90f3403 3d81c4a 67aa832 2c343af 362ebce 6a62bb3 7934af1 6a62bb3 362ebce a879a37 362ebce 3d81c4a b671eb0 3d81c4a 7e539c7 3ad179d f05adf8 3d81c4a d095de9 3d81c4a 6b284a4 5157745 3d81c4a 3d4cc61 f752342 a18612b 2608406 3c88a32 a18612b 3d81c4a |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 |
# coding=utf-8
'''DiaBLA: "Dialogue Bilingue" Bilingual dialogue dataset'''
import json
import datasets
logger = datasets.logging.get_logger(__name__)
_CITATION = '''\
@article{bawden_DiaBLa:-A-Corpus-of_2021,
author = {Bawden, Rachel and Bilinski, Eric and Lavergne, Thomas and Rosset, Sophie},
doi = {10.1007/s10579-020-09514-4},
title = {DiaBLa: A Corpus of Bilingual Spontaneous Written Dialogues for Machine Translation},
year = {2021},
journal = {Language Resources and Evaluation},
publisher = {Springer Verlag},
volume = {55},
pages = {635--660},
url = {https://hal.inria.fr/hal-03021633},
pdf = {https://hal.inria.fr/hal-03021633/file/diabla-lre-personal-formatting.pdf},
}
'''
_DESCRIPTION = '''\
English-French parallel dataset for the evaluation of \
Machine Translation (MT) for informal, written bilingual dialogue.
'''
_URLS = {
'test': 'DiaBLa.json',
}
class DiablaConfig(datasets.BuilderConfig):
'''BuilderConfig for DiaBLa.'''
def __init__(self, **kwargs):
"""BuilderConfig for DiaBLa.
Args:
**kwargs: keyword arguments forwarded to super.
"""
super(DiablaConfig, self).__init__(**kwargs)
class Diabla(datasets.GeneratorBasedBuilder):
'''DiaBLa: English-French parallel dataset of bilingual dialogue'''
BUILDER_CONFIGS = [
DiablaConfig(
name='plain_text',
version=datasets.Version('1.0.0', ''),
description='Plain text',
),
]
#TODO
def _info(self):
return datasets.DatasetInfo(
description=_DESCRIPTION,
features=datasets.Features(
{
'id': datasets.Value('string'),
'orig': datasets.Value('string'),
'norm': datasets.Value('string'),
'mt': datasets.Value('string'),
'ref': datasets.Value('string'),
'utterance_meta': {
'eval_judgment': datasets.Value("string"),
'eval_verbatim': datasets.Value('string'),
'eval_problems': [
datasets.Value("string")
],
'lang': datasets.Value("string")
},
'dialogue_meta': {
'start_time': datasets.Value('string'),
'end_time' : datasets.Value('string'),
'translation_model': datasets.Value('string'),
'final_evaluation_user1': {
'style': datasets.Value("string"),
'coherence': datasets.Value("string"),
'grammaticality': datasets.Value("string"),
'meaning': datasets.Value("string"),
'word_choice': datasets.Value("string"),
},
'final_evaluation_user2': {
'style': datasets.Value("string"),
'coherence': datasets.Value("string"),
'grammaticality': datasets.Value("string"),
'meaning': datasets.Value("string"),
'word_choice': datasets.Value("string"),
},
'scenario': [[
datasets.Value("string")
]],
'user1': {
'role_num': datasets.Value('int64'),
'role':[
datasets.Value('string')
],
'initiated_dialogue': datasets.Value('bool'),
'turn_number': datasets.Value('int64'),
'lang': datasets.Value("string"),
},
'user2':{
'role_num': datasets.Value('int64'),
'role':[
datasets.Value('string')
],
'initiated_dialogue': datasets.Value('bool'),
'turn_number': datasets.Value('int64'),
'lang': datasets.Value("string"),
}
},
'dialogue_history': [
{
'id': datasets.Value('string'),
'orig': datasets.Value('string'),
'norm': datasets.Value('string'),
'mt': datasets.Value('string'),
'ref': datasets.Value('string'),
'utterance_meta': {
'eval_judgment': datasets.Value("string"),
'eval_verbatim': datasets.Value("string"),
'eval_problems': [
datasets.Value("string")
],
'lang': datasets.Value("string"),
}
}
]
}
),
supervised_keys=None,
homepage='https://github.com/rbawden/DiaBLa-dataset',
citation=_CITATION,
)
def _split_generators(self, dl_manager):
downloaded_files = dl_manager.download_and_extract(_URLS)
return [datasets.SplitGenerator(name=datasets.Split.TEST, gen_kwargs={'filepath': downloaded_files['test']})]
def _generate_examples(self, filepath):
'''This function returns the examples in the raw (text) form.'''
logger.info("generating examples from = %s", filepath)
key = 0
with open(filepath, encoding="utf-8") as f:
diabla = json.load(f)
for dialogue_name in sorted(diabla['dialogues']):
dialogue_history = [] # to store past utterances
dialogue = diabla['dialogues'][dialogue_name]
# Meta-information attached to the dialogue
dialogue_info_keys = ['start_time', 'end_time', 'scenario',
'user1', 'user2', 'translation_model',
'final_evaluation_user1', 'final_evaluation_user2']
for user in 'user1', 'user2':
dialogue[user]['role_num'] = dialogue[user].get('role_num', dialogue[user].get('rolenum', ''))
for info_to_remove in ['eval-stage', 'useragent', 'rolenum']:
if info_to_remove in dialogue[user]:
del dialogue[user][info_to_remove]
dialogue_info = {k: dialogue[k] for k in dialogue_info_keys}
if dialogue_info['end_time'] is None:
dialogue_info['end_time'] = ''
for final_eval in 'final_evaluation_user1', 'final_evaluation_user2':
# Initialise when empty
if dialogue_info[final_eval] == {}:
dialogue_info[final_eval] = {'grammaticality': '', 'meaning': '',
'coherence': '', 'style': '', 'word_choice': ''}
# Remove some information
for info_to_remove in ['interface','verbatim_quality',
'particular_problems', 'tech',
'would_use', 'timestamp', 'technical_issue']:
if info_to_remove in dialogue_info[final_eval]:
del dialogue_info[final_eval][info_to_remove]
# Main data: the utterances
for utterance_id in dialogue['utterances']:
utterance = dialogue['utterances'][utterance_id]
# Meta-information attached to the utterance
utterance_info_keys = ['judgment', 'verbatim', 'problems']
utterance_info = {'eval_' + k: utterance['eval'][k] for k in utterance_info_keys}
if utterance_info['eval_judgment'] is None:
utterance_info['eval_judgment'] = ''
utterance_info['lang'] = utterance['language']
# Utterance text
original_text = utterance['original_text']
mt_text = utterance['postprocessed_text']
reference_text = utterance['reference_translation']
normalised_text = utterance['normalised_version']
id_ = dialogue_name + '_' + utterance_id
utterance_instance = {
'orig': original_text,
'norm': normalised_text,
'mt': mt_text,
'id': id_,
'ref': reference_text.replace('’', "'").replace('…', '...'), # normalise apostrophes to be the same as mt
'utterance_meta': utterance_info
}
# add to history (without dialogue info and history)
minimal_utterance = utterance_instance.copy()
utterance_instance['dialogue_meta'] = dialogue_info
utterance_instance['dialogue_history'] = dialogue_history.copy()
dialogue_history.append(minimal_utterance)
yield id_, utterance_instance
|