Measuring Online Hate on 4chan Using Deep Learning

This is the dataset released with the paper titled: "Measuring Online Hate on 4chan Using Deep Learning".

This dataset contains a collection of 500,000 posts extracted from the /pol/ board (Politically Incorrect) of 4chan using the 4chan API. The dataset is structured as a single CSV file with one column, com, which includes the raw content of the posts.

The dataset does not preserve the structure of threads or replies; instead, it consists of a flat collection of individual posts extracted from /pol/. This format is intended to support applications such as text analysis, natural language processing, and computational social science research by providing a straightforward dataset of raw post content.

Dataset Format

• File Format: CSV (Comma-Separated Values)
• Columns:
  • com: The raw content of the post.

Source

The posts were extracted from 4chan’s /pol/ board using the official 4chan API. This board is known for hosting discussions on various topics, often with a focus on political content. Due to the nature of the /pol/ board, the content may include offensive language, hate speech, or otherwise sensitive material. Users should exercise caution and consider ethical implications when analysing this dataset.

Potential Use Cases

• Text analysis and natural language processing (NLP).
• Studies on online discourse, extremism, or political polarization.
• Research on language usage and sentiment in online forums.
• Development and testing of machine learning models for text classification or moderation.

Example Data

Here’s an example of what a few rows of the dataset look like:

If you find our dataset useful, please cite our paper:

@article{
 
}