Developers: Integrate Your App with Gowalla for Better Friend Recommendations

Kira M. Newman Aug 10, 2011 Tweet

Downloading a new app often means facing a sparse feed, an empty inbox, and updates shared with practically no one. It’s like going to a bad party, the opposite of what social is supposed to be. But good friend recommendations—especially those based on check-ins, says new research out of the University of Cambridge’s Computer Laboratory—can spice things up.

“We are likely to become friends with ‘friends of friends,’ but what we find is there are specific places which foster the creation of new friendships,” says Salvatore Scellato, a Cambridge PhD student and coauthor of the study. Friends of friends, currently the favored approach to recommendations, account for 50% of new connections on Gowalla. But this can get unwieldy; if my hundreds of friends each have hundreds of friends, that’s tens of thousands of possible recommendations. Factoring in check-ins reduces the pool of possibilities and predicts 66 percent of new links.

Scellato and his team analyzed over 12 million check-ins made during 4 months. They assigned weights to different types of locations, with more emphasis on those with fewer total check-ins and more repeat visits—schools or offices, rather than airports or sports stadiums.

These findings are particularly relevant for startups struggling to retain new users on a social network or social app. Even Google, where Scellato is interning this summer, could learn something: Google Plus recommendations relied on email contacts, and I found myself sorting through many irrelevant and redundant suggestions.

Check-ins could also predict things other than future connections. Coauthor Anastasios Noulas is interested in recommending activities to users based on their location and the time—though “that is pretty hard,” he admits. Foursquare already uses check-in history along with your friends’ favorites to recommend places to visit, and check-ins could enhance the algorithms behind event or deal recommendation software. With users constantly digging through heaps of content, deals, and new apps, snagging them with excellent recommendations can brighten a company’s future, too.