Enterprise 2.0 Tip: Data Mine Email Archives to Generate A Social Map

One of the biggest challenges in social networks is keeping them updated. When you first log in, its a blank slate, and you have to find all your friends and make connections to them. This is a bit of a pain, so sites like Facebook and LinkedIn allow you to to import your email address book. They then data-mine the address book to see who you know that might already be in the network, which helps you make lots of connections quickly.

Ignoring the obvious security and privacy concerns, there are still two big problems with this:

  1. These systems find connections, but they ignore the strength and quality of those connections.
  2. You have to constantly import your address book if you keep making new friends.

In my latest book, I give some practical advice about how Content Management fits in with social software and Enterprise 2.0 initiatives... One of the ideas that I liked to drive home is that not all connections are equal, and it takes a lot of effort to keep quality information in your social software systems. Who is connected to whom? Which connections are genuine? And who is just a "link mooch" who is spamming people with "friend" requests just to ratchet up his ranking?

That latter one is particularly problematic on LinkedIn... Its littered with sub-par recruiters who send friend request spam so they can get something from you... but they never care to do anything for you.

Luckily, in the enterprise these problems can be solved relatively easily: data mine your email archives for who is connected to whom! By monitoring a host of statistics on who emails whom, about what, and when, you have a tremendously powerful tool for building social maps. You can determine who is connected to whom, who is an expert on which subject, and where the structural holes are in your enterprise. And you never need to maintain your connections! Any time you send a message to a friend, your social map is automatically rebuilt for you!

In order to do so, you'll need to run some data mining tools to find answers to the following questions:

  • Who do you send emails to? These are the people you claim to be connected to.
  • Does this person reply to your emails? If so, the connection is mutual.
  • How often do you email? A one-time email is probably not a connection, but a weekly email might be a strong connection.
  • How long does it take them to reply to you? A faster reply usually means your communications get priority to them, and they feel a stronger connection to you.
  • How long do you take to reply to them? Again, a faster reply from you means that their communications get priority from you, meaning you feel a strong connection as well.
  • Do you answer emails about a topic, or just forward them along? Just because you are the "point man" for Java questions, that doesn't mean you "know" Java... but it probably means you "know who knows" Java, which is sometimes even better.
  • Does one person usually do all of the initiation of new emails? If so, then this might be a lopsided friendship, or it might just mean that one person has more free time.
  • What are the topics of conversation? In reality, the more often you discuss work, the weaker the connection! If you also discuss gossip, news, current events, sports, movies, family, or trivia, then you probably have a stronger connection. The more topics you discuss, the more likely you are to be close friends.
  • What is the flow of email from one department to another? If its peer-to-peer, then these departments are comfortable sharing information. If it always goes through the chain of command, then these departments are socially isolated, and probably unlikely to trust each other.
  • Who do you email outside the company? If an employee in the marketing department emailed a friend who works at the company Ravenna, and your sales person is trying to connect with somebody at Ravenna, then these two employees might want to connect.

Unfortunately, many employers have a policy against using company email for personal communications. Ironically, this policy could hurt the employer in the long run, because analyzing the violations of that policy are frequently the best way to determine who is well connected in your company! So, before you deploy any social software in the enterprise, encourage your employees to goof off via email (within reason), and set up some technology to data-mine your email archives (like Oracle Universal Online Archive, or something similar). Then keep tuning your map based on the email messages people send.

That will help you hit the ground running with enterprise social software...

You know

Most of the time, I have absolutely no idea what you're talking about. =)

This concept would be cool to see in public applications though. Because of a couple of different games that I have played on FaceBook, I have a few hundred "friends". People I don't know, will never email, and will never know any better than a number in an alliance. It would be great to see friends ranked and then set my own criteria of what I find important for that ranking. The things you listed would be high on the list but it would be good to be able to set up other criteria. The problem of privacy becomes an issue in the personal world but I still like this concept.

I wrote about this

last year in Computer Technology Review: http://tinyurl.com/csage2
Leveraging implicit or organic social networks that emerge from other activities will is the next step.

Recent comments