A Comparison of String Distance Metrics on Usernames for Cross-Platform Identification

People often use similar usernames across different social media sites. This fact can be used to correlate accounts between different platforms. Since the first mention of this fact in 2009 no research has been done on how to exploit it most efficiently. We showed that ignoring the casing will most definitely improve the matching and we found that Smith-Waterman provides the best metric to match usernames and achieves a success rate of 76%. This implies that earlier work using other string matching metrics could achieve better results by using Smith-Waterman.