CoDACom
A test suite, referencing and normalising multiple community detection algorithms, in order to easily and quickly compare associated methods.
- Public project
- Keyword(s) : Test suite, Community detection, Quality function
- Team(s) : AMACC
- License(s) : CeCILL-B
- Language(s) : Python , Javascript, CSS, Shell, C++, PHP
- Web site : https://codacom.greyc.fr/
- System(s) : MacOS X, UNIX-like
- State : Stable
- Support : Maintained
Contributor Supervision Contract Jean Creusefond Université de Caen Basse-Normandie Contrat doctoral - https://forge.greyc.fr/projects/codacom
Main features
CoDACom is a test suite aiming at simplifying the research in community detection.
It executes multiple community detection programs, each of those producing a clustering of the input graph. Results are then analysed using different metrics, called quality functions. If a ground-truth exists, it is taken into account as a reference clustering. This process enables meta-comparison, i.e. quality function analysis.
This software has been designed for easy extension. It is simple to add one's own implementation of a community detection algorithm (no rewriting of the code is necessary), or even one's own quality function. A consequent set of quality functions and community detection programs is included.
Provided implementations are, for most of them, efficient in nowadays standards : graphs with millions of edges are analysed in a few hours using a standards desktop.
The use of standard tools for dependancy and installation enable the use on multiple Unix-based systems, and has been tested on debian and OSX. A web graphical interface is included, representing most of the functionalities.
Context
Cf. https://codacom.greyc.fr/install.php
Publications
Creusefond J., Largillier T. & Peyronnet, S. (2016, January). On the evaluation potential of quality functions in community detection for different contexts, NetSci-X 2016
