User Tools

Site Tools


algorithms:communities_louvain

Louvain Communities

Finds a network partition that maximizes the modularity using an agglomerative procedure. A community is a set of nodes with more edges inside than connections with other communities. Since the problem of modularity maximization is NP-complete, the Louvain method is not guaranteed to obtain the absolute maximum, but usually gives a good approximation to the optimal value. this method is one of the most advanced algorithms available on community detection, both for its performance and the quality of its results.

Input parameter Type Default Description
Resolution*float > 01.0Resolution parameter for the modularity (usually denoted by t), which controls the number of communities identified. Reduce it to increase the number of communities, and increase it to reduce the number of communities.
Link Weight text None Name of the link property that plays the role of strength. Its values should be positive numbers.
Seed int 0Random seed for the algorithm initialization. Set it to zero to let the system choose it randomly from the computer clock.

* Required Field


The following three pictures represent three different partitions in communities of a social network. They correspond to three resolution parameters: 0.2, 1.0 and 3.0 from left to right, reducing the total number of communities.



Similarly, with the standard resolution parameter we obtain 2 communities in the Zachary's Karate Club network, but if we reduce it to t=0.5 we obtain three different clusters.



References:

algorithms/communities_louvain.txt · Last modified: 2018/12/20 14:20 by systems