Finally, the results of a coop­er­a­tion with the guys from eigen­fac­tor are online!

For the impa­tient: here’s the direct link: http://well-formed.eigenfactor.org

The site fea­tures 4 dif­fer­ent visu­al­iza­tions, try­ing dif­fer­ent approaches to map­ping infor­ma­tion flow and cita­tion struc­ture in the sciences.

A radial visu­al­iza­tion based on hier­ar­chi­cal edge bundling does a great job of dis­play­ing the over­all cita­tion and clus­ter­ing struc­ture. I love the organic feel the bun­dled splines bring in, and mul­ti­ply­ing the line col­ors added that deep extra color twist. Aes­thet­i­cally, my clear favorite. Its down­side is that it does not dis­play the direc­tion of infor­ma­tion flow very well. Also, I am not 100% happy with the very heavy dis­play when you click a whole clus­ter, but I guess you can­not have all at the same time.

A stacked col­umn visu­al­iza­tion shows how clus­ter struc­ture and jour­nal impor­tance change over time. It does not dis­play any inter-journal cita­tion infor­ma­tion, how­ever. Rem­i­nis­cent – on first sight – of his­tory flow, or other stacked charts, we were, in fact, inspired by Sankey dia­grams, and tried to map change start­ing from the clus­ter­ing struc­ture. The truth is, 95% of the data does not change much :) but data-wise, there is a dis­tilled ver­sion of, for instance, the very inter­est­ing devel­op­ment neu­ro­science has taken over the years. It is doc­u­mented already in a paper; we might want to pub­lish a spe­cial inter­ac­tive ver­sion for this story alone…

The treemap is prob­a­bly the most func­tional of the visu­al­iza­tions. It trans­ports the clus­ter struc­ture quite well, addi­tion­ally nicely tells the story how eigen­fac­tor scores sum up to 1, and allows, on click, to get some pretty pre­cise idea about the rela­tion of the jour­nal to oth­ers. It is a good coun­ter­part to the radial visu­al­iza­tion, with their com­ple­men­tary advantages.

Finally, a map. This is prob­a­bly the most obvi­ous approach for net­work visu­al­iza­tion, but I couldn’t resist :) espe­cially since it was a 1 day action. I used a spring embed­ding algo­rithm based on con­nec­tion strength to cal­cu­late the map coor­di­nates with cytoscape. I was quite pleased with its import, lay­out and export capa­bil­i­ties. I sim­ply exported as gml, grep’ed the out­put and voila I had some coor­di­nates to import. Great tool! The dis­tor­tion lens is cus­tom coded, and just uses two dif­fer­ent lin­ear dis­tance scales. No fish­eye, since I find these harder to control.

Here is the heart of the dis­tance scal­ing code, if any­one is inter­ested: var xx:Number = lens.x; var yy:Number = lens.y; var radius:Number = 50; var scale:Number = 5; var scaledRadius:Number = radius * scale; var dist:Number, diffX:Number, diffY:Number, diffUniX:Number, diffUniY:Number; for each (var n:Node in data.group("leaves")) { diffX = (n.props.x - xx); diffY = (n.props.y - yy); dist = Math.sqrt(Math.pow(diffX, 2) + Math.pow(diffY, 2)); if(dist < radius) { n.x = xx + scale * diffX; n.y = yy + scale * diffY; } else { diffUniX = diffX / dist; diffUniY = diffY / dist; n.x = xx + diffUniX * (scaledRadius + dist - radius); n.y = yy + diffUniY * (scaledRadius + dist - radius); } }

All visu­al­iza­tions imple­mented using flare, my favorite visu­al­iza­tion frame­work. Amaz­ing work, Mr Heer!

For all visu­al­iza­tions, the data basis came from the eigen­fac­tor team, cal­cu­lat­ing both impor­tance val­ues for indi­vid­ual jour­nals, as well as group­ing them hier­ar­chi­cally accord­ing to their cita­tion flow “neigh­bor­hoods”. You can find lots more infor­ma­tion on the eigen­fac­tor site.

The coop­er­a­tion started when the eigen­fac­tor team used a cus­tomized ver­sion of the good old rela­tion browser; later, they got in touch with me to pro­duce some more visu­al­iza­tions. We started work in sum­mer 08, and after lots of scrib­bles, exper­i­ments, pro­to­types, adjust­ments and a lit­tle visit in Seat­tle autumn, we can finally present the results. I quite enjoyed the coop­er­a­tion, sci­ence and design can be an explo­sive mix :) So, thanks to Carl Bergstrom, Mar­tin Ros­vall, Ben Alt­house, and Jevin West.

Any­ways — feed­back wel­come! Btw if you want to blog about it — here are some screen­shots.