Datasets:

Languages:
Vietnamese
ArXiv:
License:
holylovenia commited on
Commit
9ec4ec7
·
verified ·
1 Parent(s): 7dd7354

Upload vihealthqa.py with huggingface_hub

Browse files
Files changed (1) hide show
  1. vihealthqa.py +157 -0
vihealthqa.py ADDED
@@ -0,0 +1,157 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # coding=utf-8
2
+ from pathlib import Path
3
+ from typing import Dict, List, Tuple
4
+
5
+ import datasets
6
+ import pandas as pd
7
+
8
+ from seacrowd.utils import schemas
9
+ from seacrowd.utils.configs import SEACrowdConfig
10
+ from seacrowd.utils.constants import Licenses, Tasks
11
+
12
+ _CITATION = """\
13
+ @InProceedings{nguyen2022viheathqa,
14
+ author="Nguyen, Nhung Thi-Hong
15
+ and Ha, Phuong Phan-Dieu
16
+ and Nguyen, Luan Thanh
17
+ and Van Nguyen, Kiet
18
+ and Nguyen, Ngan Luu-Thuy",
19
+ title="SPBERTQA: A Two-Stage Question Answering System Based on Sentence Transformers for Medical Texts",
20
+ booktitle="Knowledge Science, Engineering and Management",
21
+ year="2022",
22
+ publisher="Springer International Publishing",
23
+ address="Cham",
24
+ pages="371--382",
25
+ isbn="978-3-031-10986-7"
26
+ }
27
+ """
28
+ _DATASETNAME = "vihealthqa"
29
+ _DESCRIPTION = """\
30
+ Vietnamese Visual Question Answering (ViVQA) consist of 10328 images and 15000 question-answer
31
+ pairs in Vietnamese for evaluating Vietnamese VQA models. This dataset is built based on 10328 randomly
32
+ selected images from MS COCO dataset. The question-answer pairs were based on the COCO-QA dataset that
33
+ was automatically translated from English to Vietnamese.
34
+ """
35
+ _HOMEPAGE = "https://huggingface.co/datasets/tarudesu/ViHealthQA"
36
+ _LANGUAGES = ["vie"]
37
+ _LICENSE = Licenses.UNKNOWN.value
38
+ _PAPER_URL = "https://link.springer.com/chapter/10.1007/978-3-031-10986-7_30"
39
+ _LOCAL = False
40
+ _URLS = {
41
+ "vihealthqa": {
42
+ "train": "https://huggingface.co/datasets/tarudesu/ViHealthQA/raw/main/train.csv",
43
+ "val": "https://huggingface.co/datasets/tarudesu/ViHealthQA/raw/main/val.csv",
44
+ "test": "https://huggingface.co/datasets/tarudesu/ViHealthQA/raw/main/test.csv",
45
+ }
46
+ }
47
+ _SUPPORTED_TASKS = [Tasks.QUESTION_ANSWERING]
48
+ _SOURCE_VERSION = "1.0.0"
49
+ _SEACROWD_VERSION = "2024.06.20"
50
+
51
+
52
+ class ViHealthQADataset(datasets.GeneratorBasedBuilder):
53
+ '''
54
+ This is a SeaCrowed dataloader for dataset Vietnamese Visual Question Answering (ViVQA), which consists of 10328 images and 15000 question-answer
55
+ pairs in Vietnamese for evaluating Vietnamese VQA models.
56
+ '''
57
+ SOURCE_VERSION = datasets.Version(_SOURCE_VERSION)
58
+ SEACROWD_VERSION = datasets.Version(_SEACROWD_VERSION)
59
+
60
+ BUILDER_CONFIGS = [
61
+ SEACrowdConfig(
62
+ name=f"{_DATASETNAME}_source",
63
+ version=SOURCE_VERSION,
64
+ description=f"{_DATASETNAME} source schema",
65
+ schema="source",
66
+ subset_id=f"{_DATASETNAME}",
67
+ ),
68
+ SEACrowdConfig(
69
+ name=f"{_DATASETNAME}_seacrowd_qa",
70
+ version=SEACROWD_VERSION,
71
+ description=f"{_DATASETNAME} SEACrowd schema",
72
+ schema="seacrowd_qa",
73
+ subset_id=f"{_DATASETNAME}",
74
+ ),
75
+ ]
76
+
77
+ DEFAULT_CONFIG_NAME = f"{_DATASETNAME}_source"
78
+
79
+ def _info(self) -> datasets.DatasetInfo:
80
+
81
+ if self.config.schema == "source":
82
+ features = datasets.Features(
83
+ {
84
+ "id": datasets.Value("string"),
85
+ "question": datasets.Value("string"),
86
+ "answer": datasets.Value("string"),
87
+ "link": datasets.Value("string")
88
+ }
89
+ )
90
+ elif self.config.schema == "seacrowd_qa":
91
+ features = schemas.qa_features
92
+ features["meta"] = {"link": datasets.Value("string")}
93
+ else:
94
+ raise ValueError(f"No schema matched for {self.config.schema}")
95
+
96
+ return datasets.DatasetInfo(
97
+ description=_DESCRIPTION,
98
+ features=features,
99
+ homepage=_HOMEPAGE,
100
+ license=_LICENSE,
101
+ citation=_CITATION,
102
+ )
103
+
104
+ def _split_generators(self, dl_manager: datasets.DownloadManager) -> List[datasets.SplitGenerator]:
105
+ """Returns SplitGenerators."""
106
+ urls = _URLS["vihealthqa"]
107
+ data_dir = dl_manager.download_and_extract(urls)
108
+ return [
109
+ datasets.SplitGenerator(
110
+ name=datasets.Split.TRAIN,
111
+ gen_kwargs={
112
+ "filepath": data_dir["train"],
113
+ "split": "train",
114
+ },
115
+ ),
116
+ datasets.SplitGenerator(
117
+ name=datasets.Split.VALIDATION,
118
+ gen_kwargs={
119
+ "filepath": data_dir["val"],
120
+ "split": "val",
121
+ },
122
+ ),
123
+ datasets.SplitGenerator(
124
+ name=datasets.Split.TEST,
125
+ gen_kwargs={
126
+ "filepath": data_dir["test"],
127
+ "split": "test",
128
+ },
129
+ ),
130
+ ]
131
+
132
+ def _generate_examples(self, filepath: Path, split: str) -> Tuple[int, Dict]:
133
+ """Yields examples as (key, example) tuples."""
134
+
135
+ raw_examples = pd.read_csv(filepath)
136
+
137
+ for eid, exam in raw_examples.iterrows():
138
+ assert len(exam) == 4
139
+ exam_id, exam_quest, exam_answer, exam_link = exam
140
+
141
+ if self.config.schema == "source":
142
+ yield eid, {"id": str(exam_id), "question": exam_quest, "answer": exam_answer, "link": exam_link}
143
+
144
+ elif self.config.schema == "seacrowd_qa":
145
+ yield eid, {
146
+ "id": str(eid),
147
+ "question_id": exam_id,
148
+ "document_id": str(eid),
149
+ "question": exam_quest,
150
+ "type": None,
151
+ "choices": [],
152
+ "context": exam_link,
153
+ "answer": [exam_answer],
154
+ "meta": {
155
+ "link": exam_link,
156
+ },
157
+ }