After 5 years of work, my paper in the Proceedings of the National Academy of Sciences is finally coming out! Essentially we show that we can identify behavioral signatures associated with friendship (hanging out on Saturday nights, traveling together, etc). These signatures that have been learned from the data can subsequently be used to infer the topology of a social network without users manually labeling relationships. It's been picked up in the scientific press including Science, BBC Tech, and New Scientist. Big thanks Nokia for supporting this research - the study wouldn't have been possible without some very generous university donations.