Off the top of my head, I'd say go look at the APIs for both FB and 4sq (try say user, friends, checkins, mayorship and location) and look at what is available for user PII and friends. Compare say username and do a little fuzzy matching on friend's names.
If say FB and 4sq share say more than x common friends, chances are you have the right mapping and you can build from that. Thrown in a Bayesian filter and pretty soon you will have a good model built up. Think of the 2 APIs as two sheets of paper with directed graphs on them. If you overlay them, the mapping becomes clear pretty quickly.
Off the top of my head, I'd say go look at the APIs for both FB and 4sq (try say user, friends, checkins, mayorship and location) and look at what is available for user PII and friends. Compare say username and do a little fuzzy matching on friend's names.
If say FB and 4sq share say more than x common friends, chances are you have the right mapping and you can build from that. Thrown in a Bayesian filter and pretty soon you will have a good model built up. Think of the 2 APIs as two sheets of paper with directed graphs on them. If you overlay them, the mapping becomes clear pretty quickly.