Gephi - graph visualization's gateway drug
Timothy Perisho profile photo
Updated April 08, 2015

Gephi - graph visualization's gateway drug

Score 7 out of 101
Vetted Review
Verified User
Review Source

Software Version


Modules Used

  • Force Atlas 3D (plug-in)
  • Data Labratory Helper (plug-in)

Overall Satisfaction with Gephi

I am the only one using Gephi in my medium-sized company with a relatively large Data team. I needed a graph visualization software and, after looking briefly at a few reviews and features, I chose Gephi and love it. I'm still the only one using it at my company, though, because the application is still pretty advanced research for us.
  • Gephi's UI, built-in layout, and built-in clustering algorithms make it very easy to get started with (assuming you have some really data and real questions you want answered).
  • Gephi's force atlas layout algorithm and MCL (Louvain) clustering algorithm are both very fast, which I think is crucial for a data visualization tool, because it allows you to play with your data and do intuitive analysis.
  • Gephi's visualization is pretty (and intuitive), which helped me sell the clustering techniques to un-initiated and non-technical managers and executives.
  • Gephi has a pretty large community of support, so when I learn about other clustering algorithms/visualization techniques, it often already exists in a Gephi plug-in.
  • I (and many others) have had to expand Gephi's memory manually by experimenting with the configuration file. I'm glad it's possible, but it should be easier.
  • Gephi sometimes crashes inexplicably and loses your work, so I have developed a habit of explicitly exporting versions of my graphs as csv's, but I think this should be handled automatically in Gephi.
  • Because it is prone to crash, ideally, Gephi would help the user manage his/her use, by estimating processing and memory for very large tasks and prompting the user to confirm their requests before executing. Instead, I just tend to avoid certain functions.
R is probably stronger from a statistics, mechanics, and customization stand-point, and has some plug-ins for graph visualization, but it can still be a bit of a black box, whereas Gephi is built around graph visualization, allowing you to really play with the data, even showing you the intermediate steps of the layout algorithms, which helps you understand how Gephi is making sense of the data and how you can use it more effectively. Its clickable GUI also makes it much easier to discover its functions and its beautiful outputs makes it easy to sell ideas to the un-initiated.
I understand better what Gephi is for and I might start using other solutions to actually operationalize the techniques I've developed in Gephi. However, I will probably always use Gephi to experiment with new graph-based data, to make beautiful graph visualizations, or to test new graph analysis techniques. Because of its speed and ease of use, I think of it now as my scratchpad for graph visualization.
I would recommend Gephi for easy ramp-up and fast graph visualization, especially for SNA (Social Network Analysis) and community discovery in general. It makes it easy to experiment with new techniques and explore your data intuitively, and even intuitively evaluate some more advanced techniques like clustering. However, if the user is already pretty advanced in graph analysis or wants specific algorithms and functions, they should look to see whether those exist in Gephi plug-ins and compare it to plug-ins in R, Cytoscape, yEd and other data visualization solutions. In particular, I have found that Tulip, although, harder to learn at first, fills many of the gaps left by Gephi. Still, it is so easy to get started on Gephi that I might still recommend it to someone who knows they will need a more sophisticated product like Tulip down-the-road.

Using Gephi

1 - Data Analysis

Using Gephi

Gephi is very intuitive and the fact that it shows its process helps the user understand what's going on. However, the animation features can really slow it down and there isn't a way to shut them off. Furthermore, the failures on saving mean you frequently have to start over. These problems disrupt the workflow and can be frustrating.