InMaps is such an exciting new feature of LinkedIn – it’s occupied my thinking on Personal Networks for the last couple of days since writing my original post on the subject. DJ Patil is the Chief Scientist at LinkedIn – and seems to have been in charge of driving this project. Watch this video to see him explaining his network (and those of a couple of others) – with the advantage of a very large piece of paper!
It’s fascinating to see that this rich map has been algorithmically defined on connections – and does not use the metadata (often inconsistent) input by LinkedIn users.
Over at Flowdata (the very best place to find updates on visualisation) – DJ took time to comment on the blog post by Nathan Yau (Flowdata’s founder). He said:
One thing that we should note about the calculation is that this only uses the “graph” of connections. We don’t use any other information. I think that is one of the very powerful aspects of this visualization. For example, in my case, it identifies my wife’s networks, students, people I went to grad school with, etc. Additionally there are a couple of reasons why this was a challenge. A) Getting everything to work in the browser in a smooth way from small networks (come on Nathan you need to add some connections :-)) to larger networks. B) The ability to “process” as many user’s networks as they use the site. There are over 85M users and that requires some serious processing power. We’ll do a more extensive write up when we can and I can say I was surprised by how much compute power we had to apply to make this real.
DJ’s key points are that the “groups” of different colours are formed by connections. He also discussed the challenges of implementing this sort of visualisation to the huge LinkedIn following. Would be interested to see how the servers have performed the last couple of days.
Well done DJ – this is certainly a real help to my research on Personal Networks.