Beth Seltzer, @beth_seltzer posted a really simple recipe for visualising Shakespearian character interaction on her blog – Shakespeare Network Hack. It’s a superb tutorial for learning how to collect data, manipulate it to expose relationships and finally visualise that data using Google Fusion Tables.
One of the most notable things about this tutorial is its simplicity of delivery. There is a real art in crafting tutorials and avoiding the temptation of providing too much background information to participants just getting their feet wet. This is a true art and I Beth cracks it well with this one. It provides a nice controllable dataset, a series of actions that are manageable and straightforward and a demonstrable outcome. There’s a great sense of gratification and it sets up a participants to start then to imagine how it might apply to their own work and how they might choose to start to add the complexity back that was removed for learning. I am so very prone myself to start to interpret the results of tutorial steps in hands-on sessions. I can hear myself saying “did you see what has happened here?” And then starting to add complexity. I am reminded in this superbly written tutorial that the initial intent should be to see a relatively rapid result from simple and controlled inputs. Then start to extend.
This week I delivered a short introduction to Fusion Tables (bedevilled by technical difficulties) that was intended to give participants a brief glimpse of what Tables are useful for. In this instance we used it mainly for spatially representing data (where it shines with a built-in geoparser). During a short tangent I demonstrated how the data you have in your table can be visualised using a variety of other Google Vis tools conveniently built in to Fusion Tables. One of the experimental (everything is experimental in Fusion Tables and remains so after X years) is the network visualisation. The spatial data didn’t show too well without manipluation and being in the middle of a lecture I couldn’t backwards engineer much to answer the question from the floor. Needless to say a sharp participant was struck by the value of this particular visualisation and queried how Google needs the data to be structured to deliver results. I admitted quickly that being used to using Gephi and Polinode I suspect separated node and edges tables etc. I meant to look into this, but thanks to Beth, I don’t have to. Fusion Tables actually does a lot of the AI and although is a simplistic and not overly flexible or extensible tool, a two column source and target list gives you an immediate result. Cool!
This is a great, short, concise and superbly written tutorial. And like magic it works very well.
I had to give it a quick run through and chose Macbeth and lo and behold after 10 minutes, I have a very nice visualisation that I can start to explore for patterns.