While many big-data providers do their best to de-identify individuals from human-subject data sets, the risk of re-identification is very real.