Your Future Friends: Predicted by Where You Go and Who You Know

Researchers at Cambridge University in Britain have developed a new approach to suggesting friends on social networks that looks at the places you visit most to determine not only who your next friend will be, but where they'll be.

Social networks that hand out "You Might Like" suggestions, like Facebook and LinkedIn, find these new buddies based on the friend-of-a-friend method -- the assumption that because you know me (and presumably like me) you'll know and like my friends, too.

While these predictions can be pretty accurate, researchers discovered that when people frequent the same places, they end up friending one another there, "accounting for 30 percent of new social links," Salvatore Scellato, one of the Cambridge University researchers, told Reuters.

The study used information from Gowalla, one of the first social networking services to use geolocation data to "check-in" to businesses. These hangout locations were given weighted value based on how conducive the environment was to social interaction, or "place entropy," as Scellato described it.

For instance, an airport or a high-traffic bus stop wouldn't score huge points, but a college-town Laundromat or cozy bar (or a combination of those two -- the very definition of awesome) would.

Discovering friends through geolocation is the only way to deftly sift through the crowded party and find new friends -- friends that you could actually hang out with in real life, too.

It just so happens that this kind of reexamination of friend suggestions is in line with the future of location-based apps as Gowalla CEO Josh Williams sees it. At a SXSW panel earlier this year, Williams said that the market needs fresh ideas to remain relevant. Maybe Cambridge University's study, using Gowalla, is paving the way for the service's next big steps.

For comprehensive coverage of the Android ecosystem, visit Greenbot.com.

Subscribe to the Best of TechHive Newsletter

Comments