holylovenia
commited on
Commit
•
fca402d
1
Parent(s):
daf93fc
Upload tgl_profanity.py with huggingface_hub
Browse files- tgl_profanity.py +115 -0
tgl_profanity.py
ADDED
@@ -0,0 +1,115 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import csv
|
2 |
+
from pathlib import Path
|
3 |
+
from typing import Dict, List, Tuple
|
4 |
+
|
5 |
+
import datasets
|
6 |
+
from datasets.download.download_manager import DownloadManager
|
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 |
+
@article{galinato-etal-2023-context,
|
14 |
+
title="Context-Based Profanity Detection and Censorship Using Bidirectional Encoder Representations from Transformers",
|
15 |
+
author="Galinato, Valfrid and Amores, Lawrence and Magsino, Gino Ben and Sumawang, David Rafael",
|
16 |
+
month="jan",
|
17 |
+
year="2023"
|
18 |
+
url="https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4341604"
|
19 |
+
}
|
20 |
+
"""
|
21 |
+
|
22 |
+
_LOCAL = False
|
23 |
+
_LANGUAGES = ["tgl"]
|
24 |
+
_DATASETNAME = "tgl_profanity"
|
25 |
+
_DESCRIPTION = """\
|
26 |
+
This dataset contains 13.8k Tagalog sentences containing profane words, together
|
27 |
+
with binary labels denoting whether or not the sentence conveys profanity /
|
28 |
+
abuse / hate speech. The data was scraped from Twitter using a Python library
|
29 |
+
called SNScrape and annotated manually by a panel of native Filipino speakers.
|
30 |
+
"""
|
31 |
+
|
32 |
+
_HOMEPAGE = "https://huggingface.co/datasets/mginoben/tagalog-profanity-dataset/"
|
33 |
+
_LICENSE = Licenses.UNKNOWN.value
|
34 |
+
_SUPPORTED_TASKS = [Tasks.ABUSIVE_LANGUAGE_PREDICTION]
|
35 |
+
_SOURCE_VERSION = "1.0.0"
|
36 |
+
_SEACROWD_VERSION = "2024.06.20"
|
37 |
+
_URLS = {
|
38 |
+
"train": "https://huggingface.co/datasets/mginoben/tagalog-profanity-dataset/resolve/main/train.csv",
|
39 |
+
"val": "https://huggingface.co/datasets/mginoben/tagalog-profanity-dataset/resolve/main/val.csv",
|
40 |
+
}
|
41 |
+
|
42 |
+
|
43 |
+
class TagalogProfanityDataset(datasets.GeneratorBasedBuilder):
|
44 |
+
SOURCE_VERSION = datasets.Version(_SOURCE_VERSION)
|
45 |
+
SEACROWD_VERSION = datasets.Version(_SEACROWD_VERSION)
|
46 |
+
|
47 |
+
SEACROWD_SCHEMA_NAME = "text"
|
48 |
+
|
49 |
+
BUILDER_CONFIGS = [
|
50 |
+
SEACrowdConfig(
|
51 |
+
name=f"{_DATASETNAME}_source",
|
52 |
+
version=SOURCE_VERSION,
|
53 |
+
description=f"{_DATASETNAME} source schema",
|
54 |
+
schema="source",
|
55 |
+
subset_id=_DATASETNAME,
|
56 |
+
),
|
57 |
+
SEACrowdConfig(
|
58 |
+
name=f"{_DATASETNAME}_seacrowd_{SEACROWD_SCHEMA_NAME}",
|
59 |
+
version=SEACROWD_VERSION,
|
60 |
+
description=f"{_DATASETNAME} SEACrowd schema",
|
61 |
+
schema=f"seacrowd_{SEACROWD_SCHEMA_NAME}",
|
62 |
+
subset_id=_DATASETNAME,
|
63 |
+
),
|
64 |
+
]
|
65 |
+
|
66 |
+
DEFAULT_CONFIG_NAME = f"{_DATASETNAME}_source"
|
67 |
+
CLASS_LABELS = ["1", "0"]
|
68 |
+
|
69 |
+
def _info(self) -> datasets.DatasetInfo:
|
70 |
+
if self.config.schema == "source":
|
71 |
+
features = datasets.Features(
|
72 |
+
{
|
73 |
+
"text": datasets.Value("string"),
|
74 |
+
"label": datasets.Value("int64"),
|
75 |
+
}
|
76 |
+
)
|
77 |
+
elif self.config.schema == f"seacrowd_{self.SEACROWD_SCHEMA_NAME}":
|
78 |
+
features = schemas.text_features(label_names=self.CLASS_LABELS)
|
79 |
+
else:
|
80 |
+
raise ValueError(f"Invalid config name: {self.config.schema}")
|
81 |
+
return datasets.DatasetInfo(
|
82 |
+
description=_DESCRIPTION,
|
83 |
+
features=features,
|
84 |
+
homepage=_HOMEPAGE,
|
85 |
+
license=_LICENSE,
|
86 |
+
citation=_CITATION,
|
87 |
+
)
|
88 |
+
|
89 |
+
def _split_generators(self, dl_manager: DownloadManager) -> List[datasets.SplitGenerator]:
|
90 |
+
"""Returns SplitGenerators."""
|
91 |
+
data_files = dl_manager.download_and_extract(_URLS)
|
92 |
+
|
93 |
+
return [
|
94 |
+
datasets.SplitGenerator(
|
95 |
+
name=datasets.Split.TRAIN,
|
96 |
+
gen_kwargs={"filepath": data_files["train"]},
|
97 |
+
),
|
98 |
+
datasets.SplitGenerator(
|
99 |
+
name=datasets.Split.VALIDATION,
|
100 |
+
gen_kwargs={"filepath": data_files["val"]},
|
101 |
+
),
|
102 |
+
]
|
103 |
+
|
104 |
+
def _generate_examples(self, filepath: Path) -> Tuple[int, Dict]:
|
105 |
+
"""Yield examples as (key, example) tuples"""
|
106 |
+
with open(filepath, encoding="utf-8") as f:
|
107 |
+
csv_reader = csv.reader(f, delimiter=",")
|
108 |
+
next(csv_reader, None) # skip the headers
|
109 |
+
for idx, row in enumerate(csv_reader):
|
110 |
+
text, label = row
|
111 |
+
if self.config.schema == "source":
|
112 |
+
example = {"text": text, "label": int(label)}
|
113 |
+
elif self.config.schema == f"seacrowd_{self.SEACROWD_SCHEMA_NAME}":
|
114 |
+
example = {"id": idx, "text": text, "label": int(label)}
|
115 |
+
yield idx, example
|