Datasets:
Mintaka first upload
Browse files- mintaka.py +178 -0
- test_mintaka.py +16 -0
mintaka.py
ADDED
@@ -0,0 +1,178 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# coding=utf-8
|
2 |
+
|
3 |
+
"""Mintaka: A Complex, Natural, and Multilingual Dataset for End-to-End Question Answering"""
|
4 |
+
|
5 |
+
import json
|
6 |
+
import datasets
|
7 |
+
|
8 |
+
logger = datasets.logging.get_logger(__name__)
|
9 |
+
|
10 |
+
_DESCRIPTION = """\
|
11 |
+
Mintaka is a complex, natural, and multilingual dataset designed for experimenting with end-to-end
|
12 |
+
question-answering models. Mintaka is composed of 20,000 question-answer pairs collected in English,
|
13 |
+
annotated with Wikidata entities, and translated into Arabic, French, German, Hindi, Italian,
|
14 |
+
Japanese, Portuguese, and Spanish for a total of 180,000 samples.
|
15 |
+
Mintaka includes 8 types of complex questions, including superlative, intersection, and multi-hop questions,
|
16 |
+
which were naturally elicited from crowd workers.
|
17 |
+
"""
|
18 |
+
|
19 |
+
_CITATION = """\
|
20 |
+
@inproceedings{sen-etal-2022-mintaka,
|
21 |
+
title = "Mintaka: A Complex, Natural, and Multilingual Dataset for End-to-End Question Answering",
|
22 |
+
author = "Sen, Priyanka and
|
23 |
+
Aji, Alham Fikri and
|
24 |
+
Saffari, Amir",
|
25 |
+
booktitle = "Proceedings of the 29th International Conference on Computational Linguistics",
|
26 |
+
month = oct,
|
27 |
+
year = "2022",
|
28 |
+
address = "Gyeongju, Republic of Korea",
|
29 |
+
publisher = "International Committee on Computational Linguistics",
|
30 |
+
url = "https://aclanthology.org/2022.coling-1.138",
|
31 |
+
pages = "1604--1619"
|
32 |
+
}
|
33 |
+
"""
|
34 |
+
|
35 |
+
_LICENSE = """\
|
36 |
+
Copyright Amazon.com Inc. or its affiliates.
|
37 |
+
Attribution 4.0 International
|
38 |
+
"""
|
39 |
+
|
40 |
+
_TRAIN_URL = "https://raw.githubusercontent.com/amazon-science/mintaka/main/data/mintaka_train.json"
|
41 |
+
_DEV_URL = "https://raw.githubusercontent.com/amazon-science/mintaka/main/data/mintaka_dev.json"
|
42 |
+
_TEST_URL = "https://raw.githubusercontent.com/amazon-science/mintaka/main/data/mintaka_test.json"
|
43 |
+
|
44 |
+
|
45 |
+
_LANGUAGES = ['en', 'ar', 'de', 'ja', 'hi', 'pt', 'es', 'it', 'fr']
|
46 |
+
|
47 |
+
_ALL = "all"
|
48 |
+
|
49 |
+
class Mintaka(datasets.GeneratorBasedBuilder):
|
50 |
+
"""Mintaka: A Complex, Natural, and Multilingual Dataset for End-to-End Question Answering"""
|
51 |
+
|
52 |
+
BUILDER_CONFIGS = [
|
53 |
+
datasets.BuilderConfig(
|
54 |
+
name = name,
|
55 |
+
version = datasets.Version("1.0.0"),
|
56 |
+
description = f"Mintaka: A Complex, Natural, and Multilingual Dataset for End-to-End Question Answering for {name}",
|
57 |
+
) for name in _LANGUAGES
|
58 |
+
]
|
59 |
+
|
60 |
+
BUILDER_CONFIGS.append(datasets.BuilderConfig(
|
61 |
+
name = _ALL,
|
62 |
+
version = datasets.Version("1.0.0"),
|
63 |
+
description = f"Mintaka: A Complex, Natural, and Multilingual Dataset for End-to-End Question Answering",
|
64 |
+
))
|
65 |
+
|
66 |
+
DEFAULT_CONFIG_NAME = 'en'
|
67 |
+
|
68 |
+
def _info(self):
|
69 |
+
return datasets.DatasetInfo(
|
70 |
+
description=_DESCRIPTION,
|
71 |
+
features=datasets.Features(
|
72 |
+
{
|
73 |
+
"id": datasets.Value("string"),
|
74 |
+
"lang": datasets.Value("string"),
|
75 |
+
"question": datasets.Value("string"),
|
76 |
+
"answerText": datasets.Value("string"),
|
77 |
+
"category": datasets.Value("string"),
|
78 |
+
"complexityType": datasets.Value("string"),
|
79 |
+
"questionEntity": [{
|
80 |
+
"name": datasets.Value("string"),
|
81 |
+
"entityType": datasets.Value("string"),
|
82 |
+
"label": datasets.Value("string"),
|
83 |
+
"mention": datasets.Value("string"),
|
84 |
+
"span": [datasets.Value("int32")],
|
85 |
+
}],
|
86 |
+
"answerEntity": [{
|
87 |
+
"name": datasets.Value("string"),
|
88 |
+
"label": datasets.Value("string"),
|
89 |
+
}]
|
90 |
+
},
|
91 |
+
),
|
92 |
+
supervised_keys=None,
|
93 |
+
citation=_CITATION,
|
94 |
+
license=_LICENSE,
|
95 |
+
)
|
96 |
+
|
97 |
+
def _split_generators(self, dl_manager):
|
98 |
+
return [
|
99 |
+
datasets.SplitGenerator(
|
100 |
+
name=datasets.Split.TRAIN,
|
101 |
+
gen_kwargs={
|
102 |
+
"file": dl_manager.download_and_extract(_TRAIN_URL),
|
103 |
+
"lang": self.config.name,
|
104 |
+
}
|
105 |
+
),
|
106 |
+
datasets.SplitGenerator(
|
107 |
+
name=datasets.Split.VALIDATION,
|
108 |
+
gen_kwargs={
|
109 |
+
"file": dl_manager.download_and_extract(_DEV_URL),
|
110 |
+
"lang": self.config.name,
|
111 |
+
}
|
112 |
+
),
|
113 |
+
datasets.SplitGenerator(
|
114 |
+
name=datasets.Split.TEST,
|
115 |
+
gen_kwargs={
|
116 |
+
"file": dl_manager.download_and_extract(_TEST_URL),
|
117 |
+
"lang": self.config.name,
|
118 |
+
}
|
119 |
+
),
|
120 |
+
]
|
121 |
+
|
122 |
+
def _generate_examples(self, file, lang):
|
123 |
+
if lang == _ALL:
|
124 |
+
langs = _LANGUAGES
|
125 |
+
else:
|
126 |
+
langs = [lang]
|
127 |
+
|
128 |
+
key_ = 0
|
129 |
+
|
130 |
+
logger.info("⏳ Generating examples from = %s", ", ".join(lang))
|
131 |
+
|
132 |
+
with open(file, encoding='utf-8') as json_file:
|
133 |
+
data = json.load(json_file)
|
134 |
+
for sample in data:
|
135 |
+
for lang in langs:
|
136 |
+
questionEntity = [
|
137 |
+
{
|
138 |
+
"name": str(qe["name"]),
|
139 |
+
"entityType": qe["entityType"],
|
140 |
+
"label": qe["label"] if "label" in qe else "",
|
141 |
+
"mention": qe["mention"],
|
142 |
+
"span": qe["span"],
|
143 |
+
} for qe in sample["questionEntity"]
|
144 |
+
]
|
145 |
+
|
146 |
+
answers = []
|
147 |
+
if sample['answer']["answerType"] == "entity" and sample['answer']['answer'] is not None:
|
148 |
+
answers = sample['answer']['answer']
|
149 |
+
elif sample['answer']["answerType"] == "numerical" and "supportingEnt" in sample["answer"]:
|
150 |
+
answers = sample['answer']['supportingEnt']
|
151 |
+
|
152 |
+
def get_label(labels, lang):
|
153 |
+
if lang in labels:
|
154 |
+
return labels[lang]
|
155 |
+
if 'en' in labels:
|
156 |
+
return labels['en']
|
157 |
+
return ""
|
158 |
+
|
159 |
+
answerEntity = [
|
160 |
+
{
|
161 |
+
"name": str(ae["name"]),
|
162 |
+
"label": get_label(ae["label"], lang),
|
163 |
+
} for ae in answers
|
164 |
+
]
|
165 |
+
|
166 |
+
yield key_, {
|
167 |
+
"id": sample["id"],
|
168 |
+
"lang": lang,
|
169 |
+
"question": sample["question"] if lang == 'en' else sample['translations'][lang],
|
170 |
+
"answerText": sample["answer"]["mention"],
|
171 |
+
"category": sample["category"],
|
172 |
+
"complexityType": sample["complexityType"],
|
173 |
+
"questionEntity": questionEntity,
|
174 |
+
"answerEntity": answerEntity,
|
175 |
+
|
176 |
+
}
|
177 |
+
|
178 |
+
key_ += 1
|
test_mintaka.py
ADDED
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from datasets import load_dataset
|
2 |
+
|
3 |
+
source = "AmazonScience/mintaka"
|
4 |
+
|
5 |
+
#dataset = load_dataset(source, "all", download_mode="force_redownload")
|
6 |
+
dataset = load_dataset(source, "all")
|
7 |
+
|
8 |
+
print(dataset)
|
9 |
+
print(dataset["train"][0])
|
10 |
+
print(dataset["train"][0:10]['question'])
|
11 |
+
|
12 |
+
|
13 |
+
dataset = load_dataset(source, "en")
|
14 |
+
dataset = load_dataset(source, "ar")
|
15 |
+
|
16 |
+
|