Network Analysis

Prof. Posner’s network analysis glossary

Getting started with network analysis:

  • Start with Scott Weingart’s guide to basic principles of network analysis. Network analysis just won’t make sense unless you have some basic vocabulary.
  • You can make simple network diagrams with Google Fusion Tables and Palladio. However, they can’t be customized to any great extent and you can’t run more sophisticated computational analyses on the data. At the moment, Palladio’s main limitation is that it doesn’t allow any customization (e.g., colors) and can’t be embedded (only exported as a static image); Google Fusion Tables network graphs can be embedded, but don’t look great and are limited to 200 nodes.
  • For more complex, highly customized network graphs, many people use Gephi. There are many Gephi tutorials, but most of them are quite hard to follow. Here are two I like: Brian Sarnacki’s tutorial; Martin Grandjean’s tutorial.
  • You can also use Cytoscape and NodeXL for network analysis. I recommend Cytoscape, which has many of Gephi’s features and a more modern interface. NodeXL is useful if you’re trying to pull in a network directly from Twitter (and are using a PC); otherwise, I don’t really recommend it.
  • I have written a detailed introduction to Cytoscape.
  • Visualizing Twitter data? Try this tutorial.
  • As time goes by, my own workflow leans more heavily on R’s igraph package, plus the JavaScript library D3 to create web-based visualizations. (NetworkD3 makes it easier to create D3 network graphs, with some compromises on customization.)  This is the best igraph tutorial I’ve found.
  • Networks, like all graphical forms, have built-in assumptions and limitations. On these, see Mushon Zer-Aviv’s “If Everything is a Network, Nothing is a Network” and Alexander Galloway’s “Are Some Things Unrepresentable?”

Projects to look at: