Introduction

This repository contains the code and part of the data for the dataset DomainDemo: a dataset of domain-sharing activities among different demographic groups on Twitter.

Directory Structure

• /code: for example scripts to load the data and workflow to generate derived metrics
• /data: for data

Data access

DomainDemo contains the following versions:

• DomainDemo-multivariate: multivariate version of the dataset
• DomainDemo-univariate: univariate version of the dataset
• derived_metrics: derived metrics for the domains

All these versions are hosted on Zenodo. Due to the sensitive nature of DomainDemo-multivariate and DomainDemo-univariate, researchers interested in accessing them need to apply for access. Detailed instructions are available on the Zenodo page.

The derived_metrics is available for public access. These metrics quantify different aspects, such as localness and audience partisanship, for over 129,000 domains. For details, please refer to the derived metrics folder.

We also provide an interactive app to allow everyone to explore the data. The app is hosted on domaindemoexplorer.streamlit.app.

Citation

If you use this dataset in your research, please cite the following paper:

@misc{yang2025domaindemo,
	title={DomainDemo: a dataset of domain-sharing activities among different demographic groups on Twitter},
	author={Kai-Cheng Yang and Pranav Goel and Alexi Quintana-Mathé and Luke Horgan and Stefan D. McCabe and Nir Grinberg and Kenneth Joseph and David Lazer},
	year={2025},
	eprint={2501.09035},
	archivePrefix={arXiv},
	primaryClass={cs.SI},
	url={https://arxiv.org/abs/2501.09035},
	journal={arxiv:2501.09035}
}