Inference and visualization of community structure in attributed hypergraphs

This repository provides a statistical framework for incorporating node attribute data into the inference of community structure in a hypergraph, employing a stochastic block model called HyperNEO [1].

I provide code for the HyperNEO in Python.

I confirmed that the code runs in Python 3.9.0. See requirements.txt for the libraries used in our code.

(1) Clone this repository:

git clone git@github.com:kazuibasou/hyperneo.git

(2) Run the command at hyperneo/:

mkdir data figs

Use four empirical attributed hypergraphs publicly available.

(1) Download preprocessed_real_data.zip in the data set [2] and unzip it.

(2) Find the files workspace1.npz, workspace1_meta.csv, hospital.npz, and hospital_meta.csv in the unzipped folder.

(3) Make folders named workplace and hospital in the folder hyperneo/data/.

(4) Put the files workspace1.npz and workspace1_meta.csv in the folder hyperneo/data/workplace/.

(5) Put the files hospital.npz and hospital_meta.csv in the folder hyperneo/data/hospital/.

Download contact-high-school.zip in the data set [3] and unzip it in the folder hyperneo/data/.

Download contact-primary-school.zip in the data set [4] and unzip it in the folder hyperneo/data/.

This generates the following structure of the directory.

hyperneo/data/
├ contact-high-school/
├ contact-primary-school/
├ hospital/
└ workplace/

Run the following command at hyperneo/:

python3 main.py

Perform inference and visualization of community structure in the workplace hypergraph by default. Uncomment any one of lines 7--10 in hyperneo/main.py to change the data set.

The figures visualizing inferred community structure in the given hypergraph will be saved in hyperneo/figs/.

See hyperneo/hyperneo.ipynb for more details.

[1] Kazuki Nakajima and Takeaki Uno. Inferring community structure in attributed hypergraphs using stochastic block models. 2024. [arXiv].

[2] Nicolo Ruggeri, 2023, "Real and Synthetic data for Hypergraph Benchmarking (processed for the Hy-MMSBM generative model)", [link], Edmond, V1.

[3] Austin R. Benson. contact-high-school dataset (labeled). [link].

[4] Austin R. Benson. contact-primary-school dataset (labeled). [link].

This source code is released under the MIT License, see LICENSE.txt.

• Kazuki Nakajima (https://kazuibasou.github.io)
• nakajima@tmu.ac.jp