3

In graph theory (Louvain community detection) a special parameter called modularity shows how pronounced the resulting community structure is. In other words it can indicate whether the graph has highly pronounced communities or if all the nodes are more or less interconnected.

I looked through the existing Javascript graph analysis libraries such as Cytoscape and JSNetworkX as well as Louvain community detection library itself but they only offer the measure of clustering, which is not as good.

I also looked into the Graph Tools of Neo4J but they don't offer this either.

Anyone knows of a Javascript library that could calculate this sort of thing? Or at least any algorithm written in any other language, so it could be ported to Javascript?

Aerodynamika
  • 7,883
  • 16
  • 78
  • 137

1 Answers1

1

In the end, I decided to rewrite Louvain Javascript community detection library and to expose the modularity measure from there. It is quite a good indicator of how pronounced community structure is in the graph. If it is close to 0 most of the nodes belong to the same community, if it is higher than 0.3-0.4 the graph has a highly pronounced community structure.

https://github.com/noduslabs/jLouvain

Aerodynamika
  • 7,883
  • 16
  • 78
  • 137