Datasets:

Languages:
Vietnamese
ArXiv:
License:
holylovenia commited on
Commit
c8f972e
·
verified ·
1 Parent(s): a52b91e

Upload vispamreviews.py with huggingface_hub

Browse files
Files changed (1) hide show
  1. vispamreviews.py +179 -0
vispamreviews.py ADDED
@@ -0,0 +1,179 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # coding=utf-8
2
+ # Copyright 2022 The HuggingFace Datasets Authors and the current dataset script contributor.
3
+ #
4
+ # Licensed under the Apache License, Version 2.0 (the "License");
5
+ # you may not use this file except in compliance with the License.
6
+ # You may obtain a copy of the License at
7
+ #
8
+ # http://www.apache.org/licenses/LICENSE-2.0
9
+ #
10
+ # Unless required by applicable law or agreed to in writing, software
11
+ # distributed under the License is distributed on an "AS IS" BASIS,
12
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13
+ # See the License for the specific language governing permissions and
14
+ # limitations under the License.
15
+ import os
16
+ from typing import Dict, List, Tuple
17
+
18
+ import datasets
19
+ import pandas
20
+
21
+ from seacrowd.utils import schemas
22
+ from seacrowd.utils.configs import SEACrowdConfig
23
+ from seacrowd.utils.constants import Licenses, Tasks
24
+
25
+ _CITATION = """
26
+ @InProceedings{10.1007/978-3-031-21743-2_48,
27
+ author="Van Dinh, Co
28
+ and Luu, Son T.
29
+ and Nguyen, Anh Gia-Tuan",
30
+ editor="Nguyen, Ngoc Thanh
31
+ and Tran, Tien Khoa
32
+ and Tukayev, Ualsher
33
+ and Hong, Tzung-Pei
34
+ and Trawi{\'{n}}ski, Bogdan
35
+ and Szczerbicki, Edward",
36
+ title="Detecting Spam Reviews on Vietnamese E-Commerce Websites",
37
+ booktitle="Intelligent Information and Database Systems",
38
+ year="2022",
39
+ publisher="Springer International Publishing",
40
+ address="Cham",
41
+ pages="595--607",
42
+ abstract="The reviews of customers play an essential role in online shopping.
43
+ People often refer to reviews or comments of previous customers to decide whether
44
+ to buy a new product. Catching up with this behavior, some people create untruths and
45
+ illegitimate reviews to hoax customers about the fake quality of products. These are called
46
+ spam reviews, confusing consumers on online shopping platforms and negatively affecting online
47
+ shopping behaviors. We propose the dataset called ViSpamReviews, which has a strict annotation
48
+ procedure for detecting spam reviews on e-commerce platforms. Our dataset consists of two tasks:
49
+ the binary classification task for detecting whether a review is spam or not and the multi-class
50
+ classification task for identifying the type of spam. The PhoBERT obtained the highest results on
51
+ both tasks, 86.89%, and 72.17%, respectively, by macro average F1 score.",
52
+ isbn="978-3-031-21743-2"
53
+ }
54
+ """
55
+
56
+ _LOCAL = False
57
+ _LANGUAGES = ["vie"]
58
+ _DATASETNAME = "vispamreviews"
59
+ _DESCRIPTION = """
60
+ The dataset was collected from leading online shopping platforms in Vietnam. Some of the most recent
61
+ selling products for each product category were selected and up to 15 reviews per product were collected.
62
+ Each review was then labeled as either NO-SPAM, SPAM-1 (fake review), SPAM-2 (review on brand only), or
63
+ SPAM-3 (irrelevant content).
64
+ """
65
+
66
+ _HOMEPAGE = "https://github.com/sonlam1102/vispamdetection/"
67
+ _LICENSE = Licenses.CC_BY_NC_4_0.value
68
+ _URL = "https://raw.githubusercontent.com/sonlam1102/vispamdetection/main/dataset/vispamdetection_dataset.zip"
69
+
70
+ _Split_Path = {
71
+ "train": "dataset/train.csv",
72
+ "validation": "dataset/dev.csv",
73
+ "test": "dataset/test.csv",
74
+ }
75
+
76
+ _SUPPORTED_TASKS = [Tasks.SENTIMENT_ANALYSIS] # Text Classification
77
+ _SOURCE_VERSION = "1.0.0"
78
+ _SEACROWD_VERSION = "2024.06.20"
79
+
80
+
81
+ class ViSpamReviewsDataset(datasets.GeneratorBasedBuilder):
82
+ """
83
+ The SeaCrowd dataloader for the review dataset shopping platforms in Vietnam (ViSpamReviews).
84
+ """
85
+
86
+ CLASS_LABELS = [0, 1]
87
+ SPAM_TYPE_LABELS = [0, 1, 2, 3]
88
+
89
+ BUILDER_CONFIGS = [
90
+ SEACrowdConfig(
91
+ name=f"{_DATASETNAME}_source",
92
+ version=datasets.Version(_SOURCE_VERSION),
93
+ description=f"{_DATASETNAME} source schema",
94
+ schema="source",
95
+ subset_id=f"{_DATASETNAME}",
96
+ ),
97
+ SEACrowdConfig(
98
+ name=f"{_DATASETNAME}_spam_source",
99
+ version=datasets.Version(_SOURCE_VERSION),
100
+ description=f"{_DATASETNAME} source schema",
101
+ schema="source",
102
+ subset_id=f"{_DATASETNAME}",
103
+ ),
104
+ SEACrowdConfig(
105
+ name=f"{_DATASETNAME}_seacrowd_text",
106
+ version=datasets.Version(_SEACROWD_VERSION),
107
+ description=f"{_DATASETNAME} SEACrowd schema ",
108
+ schema="seacrowd_text",
109
+ subset_id=f"{_DATASETNAME}",
110
+ ),
111
+ SEACrowdConfig(
112
+ name=f"{_DATASETNAME}_spam_seacrowd_text",
113
+ version=datasets.Version(_SEACROWD_VERSION),
114
+ description=f"{_DATASETNAME} SEACrowd schema ",
115
+ schema="seacrowd_text",
116
+ subset_id=f"{_DATASETNAME}_spam",
117
+ ),
118
+ ]
119
+
120
+ DEFAULT_CONFIG_NAME = f"{_DATASETNAME}_source"
121
+
122
+ def _info(self) -> datasets.DatasetInfo:
123
+ if self.config.name.endswith("source"):
124
+ features = (datasets.Features
125
+ (
126
+ {"id": datasets.Value("int32"),
127
+ "text": datasets.Value("string"),
128
+ "label": datasets.Value("string"),
129
+ "spam_label": datasets.Value("string"),
130
+ "rating": datasets.Value("int32")
131
+ }
132
+ ))
133
+
134
+ elif self.config.name == "vispamreviews_seacrowd_text":
135
+ features = schemas.text_features(label_names=self.CLASS_LABELS)
136
+ elif self.config.name == "vispamreviews_spam_seacrowd_text":
137
+ features = schemas.text_features(label_names=self.SPAM_TYPE_LABELS)
138
+ else:
139
+ raise ValueError(f"Invalid schema {self.config.name}")
140
+
141
+ return datasets.DatasetInfo(
142
+ description=_DESCRIPTION,
143
+ features=features,
144
+ homepage=_HOMEPAGE,
145
+ license=_LICENSE,
146
+ citation=_CITATION,
147
+ )
148
+
149
+ def _split_generators(self, dl_manager: datasets.DownloadManager) -> List[datasets.SplitGenerator]:
150
+ file_paths = dl_manager.download_and_extract(_URL)
151
+ return [
152
+ datasets.SplitGenerator(
153
+ name=datasets.Split.TRAIN,
154
+ gen_kwargs={"filepath": os.path.join(file_paths, _Split_Path["train"])},
155
+ ),
156
+ datasets.SplitGenerator(
157
+ name=datasets.Split.VALIDATION,
158
+ gen_kwargs={"filepath": os.path.join(file_paths, _Split_Path["validation"])},
159
+ ),
160
+ datasets.SplitGenerator(
161
+ name=datasets.Split.TEST,
162
+ gen_kwargs={"filepath": os.path.join(file_paths, _Split_Path["test"])},
163
+ ),
164
+ ]
165
+
166
+ def _generate_examples(self, filepath) -> Tuple[int, Dict]:
167
+ """Yields examples as (key, example) tuples."""
168
+ data_lines = pandas.read_csv(filepath)
169
+ for rid, row in enumerate(data_lines.itertuples()):
170
+ if self.config.name.endswith("source"):
171
+ example = {"id": str(rid), "text": row.Comment, "label": row.Label, "spam_label": row.SpamLabel,
172
+ "rating": row.Rating}
173
+ elif self.config.name == "vispamreviews_seacrowd_text":
174
+ example = {"id": str(rid), "text": row.Comment, "label": row.Label}
175
+ elif self.config.name == "vispamreviews_spam_seacrowd_text":
176
+ example = {"id": str(rid), "text": row.Comment, "label": row.SpamLabel}
177
+ else:
178
+ raise ValueError(f"Invalid schema {self.config.schema}")
179
+ yield rid, example