What’s In a Name?
Who is Hajj Abdul Rahman Salih? Should a government border official permit him to enter the country? Can a U.S. bank make a loan to this person? Can he buy a gun from a U.S.-based gun dealer? Should security officers take notice if he shows up at an airport with a ticket?
Government agencies and certain private businesses make such decisions, in large part, by matching people’s names against names on terrorist and criminal watch lists generated over the years. Before now, none of this has worked very well. “The government has been using archaic software to search watch lists, and people have been misidentified,” says Dr. John Hermansen, CEO of the Herndon, Va.-based Language Analysis Systems (LAS) Inc., which has developed a new line of off-the-shelf name matching software tools now used by law enforcement, U.S. Intelligence and border inspection agencies, as well as private businesses.
The LAS line of products contains a virtual name encyclopedia, assembled from 20 years of experience with U.S. government agencies and drawing on research involving the cultural nuances and spelling variations of more than 850 million names. Over the years, government contracts with LAS have typically been sole-source awards since other companies tend to approach the problem of checking names with older techniques.
These older techniques can call out innocent people in a search, while overlooking watch-listed individuals, Hermansen says. The earliest name matching systems came into use during the 1890 census. These systems dropped vowels and worked with a key string of consonants. Using the key consonants, officials would evaluate names with the goal of eliminating duplicates. Since 1890, key-based systems have been modified and upgraded to serve the needs of watch-list matching. But the basic approach has essentially never been altered. “Key-based systems immediately eliminate most of the names in a file without evaluating them,” Hermansen says. “This risks excluding the name you might be looking for. Our approach provides information about names and then helps to evaluate all the names on a watch list.”
For example, few security officers would know much about the name Hajj Abdul Rahman Salih, a sample name used by LAS to illustrate its capabilities. According to LAS, Hajj or Haj or Hadj is a title given an individual who has made the pilgrimage to Mecca, while Abdul is a prefix meaning “servant of.” The meaningful part of this string of names, notes the software, appears in the last two names: Rahman Salih. Both of these names can be spelled in at least seven different ways. The LAS system would point out all of these facts and note the alternate spellings. Older key-based systems would likely provide none of the facts, give equal weight to all four names, and miss the spelling variations that employ different consonants, such as Abusaleh Aboulrahman.
Such failures have had dire consequences in the past. In 1993, a man calling himself Mir Amal Kasi passed through a watch-list check at the U.S. border and entered the country. He passed another name check when acquiring a driver’s license. Still another watch-list check missed him when he bought an AK-47. Then it was too late. He shot five people in front of CIA headquarters in Virginia. “He was on all the watch lists as Mir Amal Kansi,” says Hermansen. “By dropping the “n” from the spelling of his name and using a legitimate alternative spelling, he got through.”
The LAS approach would have had a better chance of stopping Kasi, Hermansen continues. The system would have identified Kasi as an Urdu name and noted Pakistan as its most likely country of origin. The software would then have searched for Urdu names and spelling variations in its self-contained database, probably coming up with Kansi. “My reaction to this would be that someone calling himself Kasi may have experience with names and border watch-lists,” Hermansen says. “Even if you had little experience, you would probably say this is worth looking into.”
In the days following the Sept. 11 terrorist attacks, LAS proved its capabilities. U.S. special agents used the LAS Name Recognition Library to find spelling variations for the hijackers. By testing the variants against watch lists, the agents tracked the terrorists from their points of entry to their connections in Florida, illustrating just how important it can be to understand what’s in a name.
