Skip to content

GiulioRossetti/cdlib

CDlib - Community Detection Library

codecov Build Documentation Status CodeQL pyversions PyPI version Anaconda-Server Badge Code style: black Downloads Downloads DOI SBD++

plot

CDlib is a meta-library for community detection in complex networks: it implements algorithms, clustering fitness functions as well as visualization facilities.

CDlib is designed around the networkx python library: however, when needed, it takes care to automatically convert (from and to) igraph object so to provide an abstraction on specific algorithm implementations to the final user.

CDlib provides a standardized input/output facilities for several Community Discovery algorithms: whenever possible, to guarantee literature coherent results, implementations of CD algorithms are inherited from their original projects (acknowledged on the documentation).

If you use CDlib as support to your research consider citing:

G. Rossetti, L. Milli, R. Cazabet. CDlib: a Python Library to Extract, Compare and Evaluate Communities from Complex Networks. Applied Network Science Journal. 2019. DOI:10.1007/s41109-019-0165-9

Tutorial and Online Environments

Check out the official tutorial to get started!

If you would like to test CDlib functionalities without installing anything on your machine consider using the preconfigured Jupyter Hub instances offered by SoBigData++.

Installation

CDlib requires python>=3.8.

To install the latest version of our library just download (or clone) the current project, open a terminal and run the following commands:

pip install -r requirements.txt
pip install -r requirements_optional.txt # (Optional) this might not work in Windows systems due to C-based dependencies.
pip install .

Alternatively use pip

pip install cdlib

or conda

conda create -n cdlib python=3.9
conda config --add channels giuliorossetti
conda config --add channels conda-forge
conda install cdlib

Optional Dependencies (pip package)

To simplify the installation process, the default installation does not include optional dependencies (e.g., graph-tool). If you need them, you can install them manually or run the following command:

pip install 'cdlib[C]'

This option, safe for *nix users, will install all those optional dependencies that require C code compilation.

pip install 'cdlib[pypi]'

This option will install all those optional dependencies that are not available on conda/conda-forge.

pip install 'cdlib[all]'

This option will install all optional dependencies accessible with the flag C and pypi.

(Advanced)

Due to some strict requirements, the installation of a subset of optional dependencies is left outside the previous procedures.

graph-tool

CDlib integrates the support for SBM models offered by graph-tool. To install it refer to the official documentation and install the conda-forge version of the package (or the deb version if in a *nix system).

ASLPAw

Since its 2.1.0 release ASLPAw relies on gmpy2 whose installation through pip is not easy to automatize due to some C dependencies. To address such issue test the following recipe:

conda install gmpy2 
pip install shuffle_graph>=2.1.0 similarity-index-of-label-graph>=2.0.1 ASLPAw>=2.1.0

In case this does not solve the issue, please refer to the official gmpy2 installation instructions.

Optional Dependencies (Conda package)

CDlib relies on a few packages not available through conda: to install them please use pip.

pip install pycombo
pip install GraphRicciCurvature

conda install gmpy2 
pip install shuffle_graph>=2.1.0 similarity-index-of-label-graph>=2.0.1 ASLPAw>=2.1.0

In case ASLPAw installation fails, please refer to the official gmpy2 installation instructions.

Collaborate with us!

CDlib is an active project, any contribution is welcome!

If you like to include your model in CDlib feel free to fork the project, open an issue and contact us.

How to contribute to this project?

Contributing is good, doing it correctly is better! Check out our rules, issue a proper pull request /bug report / feature request.

We are a welcoming community... just follow the Code of Conduct.