Datasets:
Tasks:
Text Classification
Modalities:
Text
Formats:
csv
Languages:
Russian
Size:
1K - 10K
Tags:
technical_support
License:
Upload TechnicalSupportCalls.py
Browse files- TechnicalSupportCalls.py +62 -0
TechnicalSupportCalls.py
ADDED
|
@@ -0,0 +1,62 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import csv
|
| 2 |
+
import datasets
|
| 3 |
+
from datasets.tasks import TextClassification
|
| 4 |
+
|
| 5 |
+
|
| 6 |
+
_DESCRIPTION = """\
|
| 7 |
+
Коллекция из более чем 2000 обращений в службу технической поддержки размеченная по 37 различным темам.
|
| 8 |
+
"""
|
| 9 |
+
|
| 10 |
+
_CITATION = """\
|
| 11 |
+
Отсутсвует
|
| 12 |
+
"""
|
| 13 |
+
|
| 14 |
+
_TRAIN_DOWNLOAD_URL = "https://huggingface.co/datasets/SIA86/TechnicalSupportCalls/blob/main/train.csv"
|
| 15 |
+
_TEST_DOWNLOAD_URL = "https://huggingface.co/datasets/SIA86/TechnicalSupportCalls/blob/main/test.csv"
|
| 16 |
+
|
| 17 |
+
|
| 18 |
+
class TechSuppCalls(datasets.GeneratorBasedBuilder):
|
| 19 |
+
"""Technical Support Calls topic classification dataset."""
|
| 20 |
+
|
| 21 |
+
def _info(self):
|
| 22 |
+
return datasets.DatasetInfo(
|
| 23 |
+
description=_DESCRIPTION,
|
| 24 |
+
features=datasets.Features(
|
| 25 |
+
{
|
| 26 |
+
"text": datasets.Value("string"),
|
| 27 |
+
"label": datasets.features.ClassLabel(num_classes=5, names=[
|
| 28 |
+
'Не работает почта ЕПС (единая почтовая система)',
|
| 29 |
+
'Отремонтировать МФУ (многофункциональное устройство)',
|
| 30 |
+
'Настроить почту на мобильном телефоне, смартфоне',
|
| 31 |
+
'Проблема с почтой (переполнен ящик, не отправляются письма)',
|
| 32 |
+
'Настроить принтер, МФУ (многофункциональное устройство)'
|
| 33 |
+
]),
|
| 34 |
+
}
|
| 35 |
+
),
|
| 36 |
+
homepage="https://huggingface.co/datasets/SIA86/TechnicalSupportCalls/tree/main",
|
| 37 |
+
citation=_CITATION,
|
| 38 |
+
task_templates=[TextClassification(text_column="text", label_column="label")],
|
| 39 |
+
)
|
| 40 |
+
|
| 41 |
+
def _split_generators(self, dl_manager):
|
| 42 |
+
train_path = dl_manager.download_and_extract(_TRAIN_DOWNLOAD_URL)
|
| 43 |
+
test_path = dl_manager.download_and_extract(_TEST_DOWNLOAD_URL)
|
| 44 |
+
return [
|
| 45 |
+
datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"filepath": train_path}),
|
| 46 |
+
datasets.SplitGenerator(name=datasets.Split.TEST, gen_kwargs={"filepath": test_path}),
|
| 47 |
+
]
|
| 48 |
+
|
| 49 |
+
def _generate_examples(self, filepath):
|
| 50 |
+
"""Technical Support Calls examples."""
|
| 51 |
+
with open(filepath, encoding="utf-8") as csv_file:
|
| 52 |
+
csv_reader = csv.reader(
|
| 53 |
+
csv_file, quotechar='"', delimiter=",", quoting=csv.QUOTE_ALL, skipinitialspace=True
|
| 54 |
+
)
|
| 55 |
+
for id_, row in enumerate(csv_reader):
|
| 56 |
+
label, title, description = row
|
| 57 |
+
# Original labels are [1, 2, 3, 4] ->
|
| 58 |
+
# ['World', 'Sports', 'Business', 'Sci/Tech']
|
| 59 |
+
# Re-map to [0, 1, 2, 3].
|
| 60 |
+
label = int(label) - 1
|
| 61 |
+
text = " ".join((title, description))
|
| 62 |
+
yield id_, {"text": text, "label": label}
|