What’s the connection between the Eiffel Tower and the Big Ben? How are you related to Mickey Mouse? Or Elvis Presley? Today, there’s a fun way to find out: Multimedia Lab’s new Web app Everything is Connected allows you to see how any two topics in this world connect. Choose a start topic (this might be you!) and watch an on-the-fly video that takes you to any destination topic you select. You’ll be amazed to discover how small the world we live in really is. In this post, I’ll take you behind the scenes of this fascinating app.

27 November 2012

Hint: I strongly recommend to try the app first :-)
If you’re curious already, check out Everything is Connected to find your connection with Mickey Mouse, Elvis Presley or anybody/anything else. And if you’re not feeling adventurous, watch my colleague Miel demonstrate the Web app.

A team project to excite people’s curiosity

As a researcher, it is of utmost importance not to live on an island. On the one hand, you have to check every once in a while how well your topic resonates in the real world. On the other hand, you have to make sure you don’t alienate from your colleagues who work on related but different topics. Especially in research, we sometimes think we work on totally different things, while for the outside world, our topics are so close they seem undistinguishable. At Multimedia Lab, where I’m part of the Semantic Web group, we know this feeling really well.

A few months ago, we therefore brainstormed on how all of us could work together on a single project that more or less encompassed all our different subtopics. We called it the überdemo back then, because it would showcase all of our talents together, and would do something cool with Linked Data on the Web. Suddenly, somebody suggested the Six degrees of separation as an idea to illustrate the concept behind Linked Data. The idea is that everyone on earth is approximately six steps away from any other person when connected through a chain of people that know each other.

We wanted to take this idea even further by connecting anything to anyone. For example, I’m a fan of Chromeo, whose record label is Last Gang Records, which is based in Los Angeles, where Walt Disney was born, who created Mickey Mouse. So this is how I’m connected to the world-famous cartoon character. But the real challenge is: can we determine these connections automatically and visualize them in a personalized documentary-style movie?

Making Linked Data tangible

How can you instruct a computer to find such a connection chain? Well, how would we as humans solve the question ourselves? If I had to do it, I’d probably go on Wikipedia and follow links from both sides to see where I can make the ends meet. This then gives me a chain of connections from one topic to another. Simple enough, and it turns out there’s actually a computer-friendly version of Wikipedia, called DBpedia. Since computers don’t understand English (yet), they need DBpedia to make sense out of Wikipedia’s articles. For example, here can you see Walt Disney’s entry on DBpedia. Note there’s lots of links there, which is exactly what we need.

Then, the idea is to feed all those DBpedia entries to an algorithm that’s able to quickly find paths. The nitty-gritty details are in our paper, but I’m sure you trust me if I say it’s pretty cool stuff :-) For the experts: it’s a variant of A* graph search. Then, we still have to visualize this path in a documentary. We thus search images on the Web and take videos from YouTube. Finally, we take several sentences from Wikipedia articles to talk you through the different topics. The talking is performed by a computerized voice. This is combined by a client-side JavaScript application that gives you a smooth, personalized video.

Now only one thing is missing: you. Because, unless you’re a celebrity or otherwise notable, you’re not likely to have an own Wikipedia page. But you might well be one of those billion people with a Facebook account. Therefore, we start from your Facebook profile and try to see if one of the things you like (such as bands or movies) is on Wikipedia. At that moment, we have a connection between you and a topic on Wikipedia, and then we can find a path towards the topic of your choice. In the end, we can show you a very tangible application of Linked Data, connecting you to any topic in the world solely based on the links between those topics.

We connected the world… and our own team

I must admit, it was a race against the clock, since we wanted to have the demo ready for the International Semantic Web Conference 2012. We literally worked until the very last moment to ensure that every visitor would get a seamless experience. The response was overwhelming: while we had initially set up one laptop for the demo night, we ultimately had 4 devices simultaneously ready for demo… and still, there was a queue of people waiting to give it a try!

The biggest surprise came at the closing ceremony, however, when we were handed the Best Demo Award for Everything is Connected. Needless to say, the joy amongst our team members was incredible. What started out as an initiative to make us leave our research islands became a highly successful team effort and a memory for the years to come. We’re only left with one big problem now… how can we do even better at next year’s edition? ;-)