Datasets:

Languages:
Thai
ArXiv:
License:
holylovenia commited on
Commit
a217749
1 Parent(s): 29f9b2c

Upload iapp_squad.py with huggingface_hub

Browse files
Files changed (1) hide show
  1. iapp_squad.py +128 -0
iapp_squad.py ADDED
@@ -0,0 +1,128 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # coding=utf-8
2
+ import json
3
+
4
+ import datasets
5
+
6
+ from seacrowd.utils import schemas
7
+ from seacrowd.utils.configs import SEACrowdConfig
8
+ from seacrowd.utils.constants import Licenses, Tasks
9
+
10
+ _DATASETNAME = "iapp_squad"
11
+ _CITATION = """\
12
+ @dataset
13
+ {
14
+ kobkrit_viriyayudhakorn_2021_4539916,
15
+ author = {Kobkrit Viriyayudhakorn and Charin Polpanumas},
16
+ title = {iapp_wiki_qa_squad},
17
+ month = feb,
18
+ year = 2021,
19
+ publisher = {Zenodo},
20
+ version = 1,
21
+ doi = {10.5281/zenodo.4539916},
22
+ url = {https://doi.org/10.5281/zenodo.4539916}
23
+ }
24
+ """
25
+
26
+ _DESCRIPTION = """
27
+ `iapp_wiki_qa_squad` is an extractive question answering dataset from Thai Wikipedia articles.
28
+ It is adapted from [the original iapp-wiki-qa-dataset](https://github.com/iapp-technology/iapp-wiki-qa-dataset)
29
+ to [SQuAD](https://rajpurkar.github.io/SQuAD-explorer/) format, resulting in
30
+ 5761/742/739 questions from 1529/191/192 articles.
31
+ """
32
+
33
+ _HOMEPAGE = "https://github.com/iapp-technology/iapp-wiki-qa-dataset"
34
+ _LICENSE = Licenses.MIT.value
35
+ _HF_URL = " https://huggingface.co/datasets/iapp_wiki_qa_squad"
36
+ _SUPPORTED_TASKS = [Tasks.QUESTION_ANSWERING]
37
+ _LOCAL = False
38
+ _LANGUAGES = ["tha"]
39
+ _SOURCE_VERSION = "1.0.0"
40
+ _SEACROWD_VERSION = "2024.06.20"
41
+
42
+ _URLS = {
43
+ "train": "https://raw.githubusercontent.com/iapp-technology/iapp-wiki-qa-dataset/main/squad_format/data/train.jsonl",
44
+ "validation": "https://raw.githubusercontent.com/iapp-technology/iapp-wiki-qa-dataset/main/squad_format/data/valid.jsonl",
45
+ "test": "https://raw.githubusercontent.com/iapp-technology/iapp-wiki-qa-dataset/main/squad_format/data/test.jsonl",
46
+ }
47
+
48
+
49
+ class IappWikiQASquadDataset(datasets.GeneratorBasedBuilder):
50
+ BUILDER_CONFIGS = [
51
+ SEACrowdConfig(name=f"{_DATASETNAME}_source", version=datasets.Version(_SOURCE_VERSION), description=_DESCRIPTION, subset_id=f"{_DATASETNAME}", schema="source"),
52
+ SEACrowdConfig(name=f"{_DATASETNAME}_seacrowd_qa", version=datasets.Version(_SEACROWD_VERSION), description=_DESCRIPTION, subset_id=f"{_DATASETNAME}", schema="seacrowd_qa"),
53
+ ]
54
+ DEFAULT_CONFIG_NAME = f"{_DATASETNAME}_source"
55
+
56
+ def _info(self):
57
+ if self.config.schema == "source":
58
+ features = datasets.Features(
59
+ {
60
+ "question_id": datasets.Value("string"),
61
+ "article_id": datasets.Value("string"),
62
+ "title": datasets.Value("string"),
63
+ "context": datasets.Value("string"),
64
+ "question": datasets.Value("string"),
65
+ "answers": datasets.features.Sequence(
66
+ {
67
+ "text": datasets.Value("string"),
68
+ "answer_start": datasets.Value("int32"),
69
+ "answer_end": datasets.Value("int32"),
70
+ }
71
+ ),
72
+ }
73
+ )
74
+ elif self.config.schema == "seacrowd_qa":
75
+ features = schemas.qa_features
76
+ features["meta"] = {
77
+ "answer_start": datasets.Value("int32"),
78
+ "answer_end": datasets.Value("int32"),
79
+ }
80
+ return datasets.DatasetInfo(description=_DESCRIPTION, features=features, homepage=_HOMEPAGE, citation=_CITATION, license=_LICENSE)
81
+
82
+ def _split_generators(self, dl_manager):
83
+ file_paths = dl_manager.download_and_extract(_URLS)
84
+ return [
85
+ datasets.SplitGenerator(
86
+ name=datasets.Split.TRAIN,
87
+ gen_kwargs={"filepath": file_paths["train"]},
88
+ ),
89
+ datasets.SplitGenerator(
90
+ name=datasets.Split.VALIDATION,
91
+ gen_kwargs={"filepath": file_paths["validation"]},
92
+ ),
93
+ datasets.SplitGenerator(
94
+ name=datasets.Split.TEST,
95
+ gen_kwargs={"filepath": file_paths["test"]},
96
+ ),
97
+ ]
98
+
99
+ def _generate_examples(self, filepath):
100
+ """Yields examples."""
101
+ with open(filepath, encoding="utf-8") as f:
102
+ for id_, row in enumerate(f):
103
+ data = json.loads(row)
104
+ if self.config.schema == "source":
105
+ yield id_, {
106
+ "question_id": data["question_id"],
107
+ "article_id": data["article_id"],
108
+ "title": data["title"],
109
+ "context": data["context"],
110
+ "question": data["question"],
111
+ "answers": {
112
+ "text": data["answers"]["text"],
113
+ "answer_start": data["answers"]["answer_start"],
114
+ "answer_end": data["answers"]["answer_end"],
115
+ },
116
+ }
117
+ elif self.config.schema == "seacrowd_qa":
118
+ yield id_, {
119
+ "id": id_,
120
+ "question_id": data["question_id"],
121
+ "document_id": data["article_id"],
122
+ "question": data["question"],
123
+ "type": "abstractive",
124
+ "choices": [],
125
+ "context": data["context"],
126
+ "answer": data["answers"]["text"],
127
+ "meta": {"answer_start": data["answers"]["answer_start"][0], "answer_end": data["answers"]["answer_end"][0]},
128
+ }