TwitterFollowGraph: Temporal User-Author Follow Graph

Provider
Twitter Cortex (El-Kishky et al.)
License
CC-BY-4.0 (source)
Size
3 files
Stored at
Hugging Face (https://huggingface.co/datasets/Twitter/TwitterFollowGraph)
Access
Free and public. No account with the data provider is needed.
Category
Machine Learning & AI
Source
https://arxiv.org/abs/2205.06205

About TwitterFollowGraph: Temporal User-Author Follow Graph

A large anonymised bipartite graph of Twitter follow engagements released by Twitter with the paper kNN-Embed: Locally Smoothed Embedding Mixtures for Multi-interest Candidate Retrieval. Consumer (user) nodes link to producer (author) nodes, and each directed edge records a user following an author. Edges are binned into contiguous time chunks denoted by ordinals that respect time order, so the graph supports temporal train and test splits. The file is a single CSV with three columns, user_index, author_index and time_chunk, distributed as TwitterFollowGraph.csv.zip (about 2.3 GB compressed, measured 2026-11-14); the repository uses about 12.6 GB. Uses include candidate retrieval and recommendation, link prediction, multi-interest user embedding, bipartite graph representation learning and benchmarking scalable embedding methods on a real social graph. Unzip and read with pandas, polars or DuckDB, then build sparse matrices with scipy or graphs with PyTorch Geometric or DGL. Released under CC BY 4.0; cite El-Kishky et al. (2022).

How to load TwitterFollowGraph: Temporal User-Author Follow Graph in Python

from huggingface_hub import hf_hub_download
path = hf_hub_download(repo_id="Twitter/TwitterFollowGraph", filename="TwitterFollowGraph.csv.zip", repo_type="dataset")
print(path)

Data structure

Files

Example file: TwitterFollowGraph.csv.zip

Analyze this dataset without downloading it

On Scigantic, TwitterFollowGraph: Temporal User-Author Follow Graph opens in a hosted JupyterLab notebook with the data already mounted, so you can read files where they are stored instead of copying them first. Each notebook includes an AI assistant that can see the mounted files and write and run analysis code. Open the dataset page to launch a notebook, or browse the full catalog.

This page is also available as Markdown and as JSON.

Related Machine Learning & AI datasets