---
dataset_name: Fake News Dataset
language:
  - en
license: gpl
tags:
  - fake-news
  - text-classification
  - nlp
size_categories:
- 0<n<10K
task_categories:
  - text-classification
---

# Fake News Dataset

## Dataset Description

From, Rada Mihalcea, the dataset curator's website: 'A dataset of fake and legitimate news, covering several domains (technology, education, business, sports, politics, entertainment and celebrity news). It consists of nearly 1,000 news, split evenly between fake and legitimate, collected through crowdsourcing or from web sources.' This is the fakeNewsDataset portion of the data-set. A celebrityDataset will be added separately.

### Source Data
The original data-set can be downloaded as a Zip from [Rada Milhacea's website](https://web.eecs.umich.edu/~mihalcea/downloads.html#FakeNews). See the paper cited below for more information.

## Dataset Structure

### Data Fields

**text**: Value(dtype='string', id=None)   
**label**: ClassLabel(names=['legit', 'fake'], id=None)  

### Data Splits

**train**: 480 examples  

### Dataset Author/Creator

Rada Mihalcea

### Licensing Information

This data is licensed by its author under the [GNU General Public License](http://www.gnu.ai.mit.edu/copyleft/gpl.html).

### Citation Information

If you use this dataset, please cite:

Veronica Perez-Rosas, Bennett Kleinberg, Alexandra Lefevre, Rada Mihalcea, Automatic Detection of Fake News, in Proceedings of the International Conference on Computational Linguistics (COLING 2018), New Mexico, NM, August 2018. https://web.eecs.umich.edu/~mihalcea/papers/perezrosas.coling18.pdf