Graph database and Social Network Analysis metrics

Question

Just wanted to ask whether graph databases like OrientDB, Neo4j and Titan are capable of computing Social Network Analysis (SNA) metrics?

I have heard that Neo4j can be used with JUNG but I don't understand where are the benefits of using graph database come in if JUNG is the only solution.

Any clarification on how metrics such as network centrality, clustering coefficient and graph components are computed would be much appreciated!

Sample code in Gremlin or Cypher would be most welcomed!

Answer 1

Although the question is already dated, here is my bit: it is possible to link Neo4j with Gephi via the Graph Streaming plugin of Gephi (it uses the GraphStreaming API). The whole process is explained on a tutorial that was submitted to The Programming Historian.

Answer 2

I recently created a GitHub project for Social Network Analytics based on Neo4j: https://github.com/besil/Neo4jSNA.

Another very interesting project is https://github.com/kbastani/neo4j-mazerunner, which uses Spark + GraphX for analysis and Neo4j as database. I have no experience with the latter, but it seems very promising.

Answer 3

TinkerPop3 will introduce the concept of GraphComputers. Read more about it here: [TinkerPop3] Lapin' up GraphComputer RFC

All the links in this post are already outdated, but that just shows that the TinkerPop team is working hard on the next release.

Cheers, Daniel

Answer 4

It is possible to export Neo4j to GraphML and to import that into NodeXL to do SNA. See the following example:

http://nodexlgraphgallery.org/Pages/Graph.aspx?graphID=16350

You can download NodeXL here: http://socialnetimporter.codeplex.com/

I am working on creating an open source importer directly from Neo4j to NodeXL. I hope to release that soon. I'll update this answer when that is ready.