HTTPS SSH

JGraphT, Social Network Analysis addon

This library implements several algorithms used in social network analysis (SNA) on top of the JGraphT library http://www.jgrapht.org/. JGraphT is a Java library for Graphs. It supports generics, where nodes and edges can be of user-defined types. The intention is to support as many SNA-specific algorithms as possible.

Our focus so far has been the following:

  1. Centrality measures: Degree centrality, closeness centrality, betweenness centralty, eigenvector centrality, etc.
  2. Hierarchy detection algorithms.
  3. Improved support for GraphML import/export (weighted and directed).
  4. Keyplayers algorithms: KPP-POS, KPP-NEG.
  5. Random-error generators and generators for complex network types.
  6. Edge prediction algorithms: Common neighbors, Adamic/Adar's measure.

Generally, the algorithms are constructed in such a way that they take the types of graphs for which makes sense to them as input. For example WeightedDegreeCentrality expects a WeightedGraph as input.

Getting started

JGraphT-SNA is built with Maven, and you can include it by including its artifact in your Maven project.

<dependency>
   <groupId>dk.aaue.sna</groupId>
   <artifactId>jgrapht-sna</artifactId>
   <version>1.3</version>
</dependency>

Artifacts are published to conjars.org:

<repository>
    <id>conjars.org</id>
    <url>http://conjars.org/repo</url>
</repository>

Alternatively, you can clone the source-code and build it yourself.

Example code

Suppose you have a graph G and you want to know its entropy centrality, you can get it like this:

Graph<String, Defaultedge> G = buildGraph();
CentralityMeasure<String> cm = new dk.aaue.sna.alg.centrality.OrtizArroyoEntropyCentrality(G);
CentralityResult<String> cr = cm.calculate();

// assuming your graph has a node "me"
System.out.println("centrality of 'me' = " + cr.get("me"));
System.out.println("rank of 'me'       = " + cr.getSortedNodes().indexOf("me"));

About

Currently, all code is written by me. Most of the code was done for my masters thesis, Aalborg University, Denmark. Feel free to use and modify the code. If anything useful, send me a pull request. Cadeau to the brilliantly smart authors of all the algorithms implemented here.

License

License is FreeBSD-style. Note JGraphT is GPL-2.1.