Figure 1

Inferring name for gender recognition. Due to some names being used for both males and females, we assign a probability of being male to each candidate based on the fraction of times their first name was assigned to a male baby in the name dataset. We define gender probability cutoffs of 0.1 and 0.9 consistent with previous studies [38]. Our gender recognition yielded 11.87% females and 88.13% males out of all users with names. All in all we found 194,010 females, 1,441,130 males, and 6,163,370 unknowns