Where You Go Is Who You Are: Using location-based social-networking services to create the next killer context

In 2008, I was struck by a comment at the Where 2.0 Conference in San Jose: “Location is a killer context.” We’re still beginning to understand how we can use this new context to build better applications. With the rise of location-based social networking services like Foursquare, the idea of “where you are” at a particular point in time has attracted much attention. But a far richer context exists and is readily available to developers. The collection of places an individual has visited provides insight into the personality and preferences of the individual.

In the past 15 years, we’ve seen the rise of incredibly effective “recommendation engines” on Netflix, Amazon, Pandora, and other sites. Given the products, movies, and music you’re interested in, it’s possible to provide the user with a recommendation of other items they will like. The same applies to location-based contexts. Using the places you go it’s possible to inform a user to other places they will enjoy and, in turn, what kind of person they are.

Imagine this scenario: it’s Saturday night and you’d like to get a drink with your friends at a nearby bar. With location-based services, it’s easy enough to use any number of mobile apps and determine what watering holes are nearby. Depending on the app, you also might be able to view the ratings for the nearby bars. But that doesn’t guarantee you’d personally enjoy any of these unfamiliar establishments – the location of the bars is certainly valuable, as is the opinion of other individuals who have visited the bar. But a critical piece of information is ignored in these mobile, location-based apps: what kind of places you’ve been in the past and what that says about you.

It’s possible to come up with an algorithm that looks at the places you’ve visited and other people have visited and come up with a decent suggestion of a new location to visit. That’s a classic recommendation-engine algorithm. We can take a shortcut, though, using already-existing tools: the Foursquare API, which provides a history of where the user has been recently, and the SpotRank API, which contains information about the popularity of a geographical area at particular times. By combining SpotRank with the locations the user has visited in the past, we can easily determine how often the individual visits places that are highly trafficked and popular relative to other places in that metropolitan area. Among many other uses, this provides an incredibly rich context to applications that find nearby restaurants and bars. For instance, when the user checks-in to a bar or restaurant, is that location getting busier or is it slowing down? During peak times, is the individual likely to retreat to quiet neighborhood haunts or mingle with the masses downtown? The previous places the user has visited combined with the relative popularity of these locations at a particular time gives us valuable information about an individual’s nightlife preferences. We can use this information to make better suggestions.

We’re just at the beginning of coming up with interesting applications that use location-based data. However, we have reached the point where existing mobile applications can begin to provide an even richer context than “you are here.” It’s particularly important to look intelligently at information related to where we’ve been. In the example above, it provides us with previously unknown information about the user than can be used to build richer location-based applications.