metadata
language:
- th
dataset_info:
features:
- name: text
dtype: string
- name: ข้อความ
dtype: string
- name: label
dtype:
class_label:
names:
'0': World
'1': Sports
'2': Business
'3': Sci/Tech
หมวด ข่าว ภาษาไทย
ตั้งต้นจาก https://huggingface.co/datasets/fancyzhx/ag_news
label
- 0 World
- 1 Sport
- 2 Business
- 3 Sci/Tech
ยังแปลไม่ครบ แต่มีแปลแล้วประมาณหนึ่ง
- train.csv: แปลแล้ว 17,271 records จาก 120,000 records
- test.csv: แปลทั้งหมดแล้ว 7,600 records
ตัวอย่างการใช้งาน
from datasets import load_dataset
# Specify the data files
data_files = {
"test": "test.csv",
"train": "train.csv"
}
dataset = load_dataset("uisp/ag_news_th", data_files=data_files)
print("Keys in loaded dataset:", dataset.keys()) # Should show keys for splits, like {'test', 'train'}
# Convert a split to pandas for further processing
test = dataset['test'].to_pandas()
print(test.head())
print(len(test.index))
train = dataset['train'].to_pandas()
train.dropna(inplace=True) # เอาตัวที่ยังไม่ได้แปล (column ข้อความ ว่างเปล่า) ออกไป
print(train.head())
print(len(train.index))```