You know who you talk to and who talks to you but what does that look like in a visualization? If you ever wondered such a thing check out MIT's Immersion project which, with your permission, takes your Gmail or MS Exchange account and analyzes your messages to figure out the who and when of your messaging.
The service only accesses the From, To, Cc and Timestamp fields of messages igand ignores the subject, body, and other content. When you log out of Immersion you can either save or delete your data which will consist of the compressed email metadata and your user profile. When this data is saved it is in a (unspecified) secure system and you can delete it whenever you like. You can also take snapshots of your graph with, or without, labels. These snapshots are saved and available for 30 days.
The visualization allows you to focus on either a single node to see just the first level of connections to that node or show all nodes and display the graph for any time window. You can also drag a node and reposition it to make the display less cluttered.
Here's my total network for the 13.4 years of email messages stored in my Gmail accoun. This amounts to 216,507 emails with 1,273 collaborators.
This is just the first release so features such as being able to see secondary and tertiary connections when you're focussed on a single node or see the chain that connects two nodes aren't available yet. I've also noticed that occasionally I'll find a relationship in the graph that is inexplicable which I'd like to understand but going through my email archive to figure out how it came about turns out to be a lot of work.
I'd love to run Immersion on the content of a corporate server and combine multiple accounts to get a corporate relationships overview. If you try Immersion out and particularly if you do so on corporate data, let me know what you think and what you find out.