|
|
--- |
|
|
license: mit |
|
|
task_categories: |
|
|
- text-classification |
|
|
language: |
|
|
- en |
|
|
tags: |
|
|
- fake-news |
|
|
- real-news |
|
|
- binary-classification |
|
|
- transformers |
|
|
pretty_name: Fake New News Detection Dataset |
|
|
size_categories: |
|
|
- 10K<n<100K |
|
|
--- |
|
|
# NewsFineTuning |
|
|
|
|
|
A combined dataset of fake and true news articles for binary classification. |
|
|
|
|
|
## Files |
|
|
|
|
|
- `train.csv` – 70% training data |
|
|
- `val.csv` – 15% validation data |
|
|
- `test.csv` – 15% test data |
|
|
|
|
|
Each CSV contains: |
|
|
- `title` — Article title |
|
|
- `text` — Full article content |
|
|
- `subject` — Topic category |
|
|
- `date` — Published date |
|
|
- `label` — `0` = Real, `1` = Fake |
|
|
|
|
|
## Task |
|
|
|
|
|
This dataset is ideal for: |
|
|
- Fake news detection |
|
|
- Binary classification |
|
|
- Fine-tuning transformers (e.g., DistilBERT) |
|
|
|
|
|
## Load with Datasets |
|
|
|
|
|
```python |
|
|
from datasets import load_dataset |
|
|
|
|
|
dataset = load_dataset("declan101/NewsFineTuning") |
