OFAC (and other sanctions lists) often provide multiple aliases for the people they list. Providing aliases for a person is essential when they operates under many identities, however many of the aliases are only slightly different presentations of the same name. Some cultures lend themselves to this, for instance Korean names can have their two given names combined, separated or hyphen separated, and have multiple transliterations of the same original character; Spanish names can appear either with or without the maternal surname.
In the graph above we can see that the number of aliases provided varies a lot by program, the program with the most entities – Counter Narcotics – does not have the most number of names to screen, that would be Counter Terrorism, due to the number of aliases provided by that program.
In many cases the differences in the aliases are within the capabilities of fuzzy matching, so we see systems trying to optimise our investigation work by only selecting the closest alias. The person with the most aliases – Abdelmalek Droukdel (47 aliases) – has 19 different aliases which only differ by a tiny amount from each other, having to investigate 19 matches to the same person would be arduous.
The balance of providing a sufficiency of aliases for screening purposes, while not overwhelming systems, is a moving target as screening systems get more sophisticated. Attacking the best alias for investigation is surely a goal of the machine learning algorithms that are touted to overhaul the investigation industry.
See more like this.