Datasets:
need support v1.0 also
Browse files
JAQKET.py
CHANGED
@@ -1,7 +1,4 @@
|
|
1 |
import json
|
2 |
-
import random
|
3 |
-
import string
|
4 |
-
import warnings
|
5 |
from typing import Dict, List, Optional, Union
|
6 |
|
7 |
import datasets as ds
|
@@ -10,7 +7,7 @@ import pandas as pd
|
|
10 |
_CITATION = """
|
11 |
@InProceedings{Kurihara_nlp2020,
|
12 |
author = "鈴木正敏 and 鈴木潤 and 松田耕史 and ⻄田京介 and 井之上直也",
|
13 |
-
title = "JAQKET:
|
14 |
booktitle = "言語処理学会第26回年次大会",
|
15 |
year = "2020",
|
16 |
url = "https://www.anlp.jp/proceedings/annual_meeting/2020/pdf_dir/P2-24.pdf"
|
@@ -28,16 +25,47 @@ This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 Intern
|
|
28 |
"""
|
29 |
|
30 |
_DESCRIPTION_CONFIGS = {
|
|
|
31 |
"v2.0": "v2.0",
|
32 |
}
|
33 |
|
34 |
_URLS = {
|
|
|
|
|
|
|
|
|
|
|
35 |
"v2.0": {
|
36 |
"train": "https://huggingface.co/datasets/kumapo/JAQKET/resolve/main/train_jaqket_59.350.json",
|
37 |
"valid": "https://huggingface.co/datasets/kumapo/JAQKET/resolve/main/dev_jaqket_59.350.json",
|
38 |
},
|
39 |
}
|
40 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
41 |
def dataset_info_v2() -> ds.Features:
|
42 |
features = ds.Features(
|
43 |
{
|
@@ -66,8 +94,13 @@ def dataset_info_v2() -> ds.Features:
|
|
66 |
|
67 |
|
68 |
class JAQKET(ds.GeneratorBasedBuilder):
|
69 |
-
VERSION = ds.Version("0.
|
70 |
BUILDER_CONFIGS = [
|
|
|
|
|
|
|
|
|
|
|
71 |
ds.BuilderConfig(
|
72 |
name="v2.0",
|
73 |
version=VERSION,
|
@@ -76,28 +109,85 @@ class JAQKET(ds.GeneratorBasedBuilder):
|
|
76 |
]
|
77 |
|
78 |
def _info(self) -> ds.DatasetInfo:
|
79 |
-
if self.config.name == "
|
|
|
|
|
80 |
return dataset_info_v2()
|
81 |
else:
|
82 |
raise ValueError(f"Invalid config name: {self.config.name}")
|
83 |
|
84 |
def _split_generators(self, dl_manager: ds.DownloadManager):
|
85 |
file_paths = dl_manager.download_and_extract(_URLS[self.config.name])
|
86 |
-
|
87 |
-
|
88 |
-
|
89 |
-
|
90 |
-
|
91 |
-
|
92 |
-
|
93 |
-
|
94 |
-
|
95 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
96 |
|
97 |
-
def
|
98 |
self,
|
99 |
file_path: Optional[str] = None,
|
100 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
101 |
):
|
102 |
if file_path is None:
|
103 |
raise ValueError(f"Invalid argument for {self.config.name}")
|
@@ -130,3 +220,16 @@ class JAQKET(ds.GeneratorBasedBuilder):
|
|
130 |
"ctxs": ctxs
|
131 |
}
|
132 |
yield q_id, example_dict
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
import json
|
|
|
|
|
|
|
2 |
from typing import Dict, List, Optional, Union
|
3 |
|
4 |
import datasets as ds
|
|
|
7 |
_CITATION = """
|
8 |
@InProceedings{Kurihara_nlp2020,
|
9 |
author = "鈴木正敏 and 鈴木潤 and 松田耕史 and ⻄田京介 and 井之上直也",
|
10 |
+
title = "JAQKET: クイズを題材にした日本語 QA データセットの構築",
|
11 |
booktitle = "言語処理学会第26回年次大会",
|
12 |
year = "2020",
|
13 |
url = "https://www.anlp.jp/proceedings/annual_meeting/2020/pdf_dir/P2-24.pdf"
|
|
|
25 |
"""
|
26 |
|
27 |
_DESCRIPTION_CONFIGS = {
|
28 |
+
"v1.0": "v1.0",
|
29 |
"v2.0": "v2.0",
|
30 |
}
|
31 |
|
32 |
_URLS = {
|
33 |
+
"v1.0": {
|
34 |
+
"train": "https://jaqket.s3.ap-northeast-1.amazonaws.com/data/aio_01/train_questions.json",
|
35 |
+
"valid": "https://jaqket.s3.ap-northeast-1.amazonaws.com/data/aio_01/dev1_questions.json",
|
36 |
+
"candidate_entities": "https://jaqket.s3.ap-northeast-1.amazonaws.com/data/aio_01/candidate_entities.json.gz",
|
37 |
+
},
|
38 |
"v2.0": {
|
39 |
"train": "https://huggingface.co/datasets/kumapo/JAQKET/resolve/main/train_jaqket_59.350.json",
|
40 |
"valid": "https://huggingface.co/datasets/kumapo/JAQKET/resolve/main/dev_jaqket_59.350.json",
|
41 |
},
|
42 |
}
|
43 |
|
44 |
+
|
45 |
+
def dataset_info_v1() -> ds.Features:
|
46 |
+
features = ds.Features(
|
47 |
+
{
|
48 |
+
"qid": ds.Value("string"),
|
49 |
+
"question": ds.Value("string"),
|
50 |
+
"answer_entity": ds.Value("string"),
|
51 |
+
"label": ds.Value("int32"),
|
52 |
+
"answer_candidates": ds.Sequence(
|
53 |
+
ds.Value("string"),
|
54 |
+
),
|
55 |
+
"contexts": ds.Sequence(
|
56 |
+
ds.Value("string")
|
57 |
+
)
|
58 |
+
}
|
59 |
+
)
|
60 |
+
return ds.DatasetInfo(
|
61 |
+
description=_DESCRIPTION,
|
62 |
+
citation=_CITATION,
|
63 |
+
homepage=_HOMEPAGE,
|
64 |
+
license=_LICENSE,
|
65 |
+
features=features,
|
66 |
+
)
|
67 |
+
|
68 |
+
|
69 |
def dataset_info_v2() -> ds.Features:
|
70 |
features = ds.Features(
|
71 |
{
|
|
|
94 |
|
95 |
|
96 |
class JAQKET(ds.GeneratorBasedBuilder):
|
97 |
+
VERSION = ds.Version("0.2.0")
|
98 |
BUILDER_CONFIGS = [
|
99 |
+
ds.BuilderConfig(
|
100 |
+
name="v1.0",
|
101 |
+
version=VERSION,
|
102 |
+
description=_DESCRIPTION_CONFIGS["v1.0"],
|
103 |
+
),
|
104 |
ds.BuilderConfig(
|
105 |
name="v2.0",
|
106 |
version=VERSION,
|
|
|
109 |
]
|
110 |
|
111 |
def _info(self) -> ds.DatasetInfo:
|
112 |
+
if self.config.name == "v1.0":
|
113 |
+
return dataset_info_v1()
|
114 |
+
elif self.config.name == "v2.0":
|
115 |
return dataset_info_v2()
|
116 |
else:
|
117 |
raise ValueError(f"Invalid config name: {self.config.name}")
|
118 |
|
119 |
def _split_generators(self, dl_manager: ds.DownloadManager):
|
120 |
file_paths = dl_manager.download_and_extract(_URLS[self.config.name])
|
121 |
+
if self.config.name == "v1.0":
|
122 |
+
return [
|
123 |
+
ds.SplitGenerator(
|
124 |
+
name=ds.Split.TRAIN,
|
125 |
+
gen_kwargs={"file_path": file_paths["train"], "entities_file_path": file_paths["candidate_entities"]},
|
126 |
+
),
|
127 |
+
ds.SplitGenerator(
|
128 |
+
name=ds.Split.VALIDATION,
|
129 |
+
gen_kwargs={"file_path": file_paths["valid"], "entities_file_path": file_paths["candidate_entities"]},
|
130 |
+
),
|
131 |
+
]
|
132 |
+
elif self.config.name == "v2.0":
|
133 |
+
return [
|
134 |
+
ds.SplitGenerator(
|
135 |
+
name=ds.Split.TRAIN,
|
136 |
+
gen_kwargs={"file_path": file_paths["train"]},
|
137 |
+
),
|
138 |
+
ds.SplitGenerator(
|
139 |
+
name=ds.Split.VALIDATION,
|
140 |
+
gen_kwargs={"file_path": file_paths["valid"]},
|
141 |
+
),
|
142 |
+
]
|
143 |
+
else:
|
144 |
+
raise ValueError(f"Invalid config name: {self.config.name}")
|
145 |
|
146 |
+
def _generate_examples_v1(
|
147 |
self,
|
148 |
file_path: Optional[str] = None,
|
149 |
+
entities_file_path: Optional[str] = None,
|
150 |
+
num_candidates: Optional[int] = None,
|
151 |
+
):
|
152 |
+
if file_path is None or entities_file_path is None:
|
153 |
+
raise ValueError(f"Invalid argument for {self.config.name}")
|
154 |
+
|
155 |
+
entities = dict()
|
156 |
+
with open(entities_file_path, "r", encoding="utf-8") as fin:
|
157 |
+
lines = fin.readlines()
|
158 |
+
for line in lines:
|
159 |
+
entity = json.loads(line.strip())
|
160 |
+
entities[entity["title"]] = entity["text"]
|
161 |
+
|
162 |
+
with open(file_path, "r", encoding="utf-8") as fin:
|
163 |
+
lines = fin.readlines()
|
164 |
+
for line in lines:
|
165 |
+
data_raw = json.loads(line.strip("\n"))
|
166 |
+
q_id = data_raw["qid"]
|
167 |
+
question = data_raw["question"].replace("_", "")
|
168 |
+
answer_entity = data_raw["answer_entity"]
|
169 |
+
answer_candidates = data_raw["answer_candidates"][:num_candidates]
|
170 |
+
|
171 |
+
if answer_entity not in answer_candidates:
|
172 |
+
continue
|
173 |
+
if len(answer_candidates) != num_candidates:
|
174 |
+
continue
|
175 |
+
|
176 |
+
contexts = [entities[answer_candidates[i]] for i in range(num_candidates)]
|
177 |
+
label = str(answer_candidates.index(answer_entity))
|
178 |
+
example_dict = {
|
179 |
+
"qid": q_id,
|
180 |
+
"question": question,
|
181 |
+
"answer_entity": answer_entity,
|
182 |
+
"label": label,
|
183 |
+
"answer_candidates": answer_candidates,
|
184 |
+
"contexts": contexts,
|
185 |
+
}
|
186 |
+
yield q_id, example_dict
|
187 |
+
|
188 |
+
def _generate_examples_v2(
|
189 |
+
self,
|
190 |
+
file_path: Optional[str] = None
|
191 |
):
|
192 |
if file_path is None:
|
193 |
raise ValueError(f"Invalid argument for {self.config.name}")
|
|
|
220 |
"ctxs": ctxs
|
221 |
}
|
222 |
yield q_id, example_dict
|
223 |
+
|
224 |
+
def _generate_examples(
|
225 |
+
self,
|
226 |
+
file_path: Optional[str] = None,
|
227 |
+
entities_file_path: Optional[str] = None,
|
228 |
+
num_candidates: int = 5,
|
229 |
+
):
|
230 |
+
if self.config.name == "v1.0":
|
231 |
+
yield from self._generate_examples_v1(file_path, entities_file_path, num_candidates)
|
232 |
+
elif self.config.name == "v2.0":
|
233 |
+
yield from self._generate_examples_v2(file_path)
|
234 |
+
else:
|
235 |
+
raise ValueError(f"Invalid config name: {self.config.name}")
|