Profiled: Language Analysis Systems
Chairman/Co-Founder/CEO: Dr. John Hermansen
No. Employees: 20
The assassinations of Frank Darling and Lansing Bennett could have been prevented. The Central Intelligence Agency employees were waiting in traffic outside headquarters in Langley, Virginia, when they were gunned down nearly 11 years ago by Mir Aimal Kansi, a Pakistani militant who entered the U.S. in December 1990 on a one-month business visa under the family name Kasi. After living in the U.S. illegally for nearly a year, Kansi applied for asylum in February 1992, murdered Darling and Bennett on January 25, 1993 using an assault rifle he'd purchased in the U.S., and fled to Pakistan the next day.
As Steven Camarota, director of research at Washington, DC's Center for Immigration Studies, notes in his white paper, "The Open Door…," Kansi was apprehended in Pakistan in June 1997 and sentenced to death for his crime in January 1998. Before his execution by legal injection on November 14, 2002, Kansi told the BBC that the attack had been a protest of America's "anti-Muslim policy in the Middle East" and that he did not regret the murders.
What's In a Name?
So, how could all of this been prevented? "Kansi's name was on all of the watch lists at the borders and consulates issuing visas, but it was entered as ‘Kansi' and he was spelling it without the ‘n'—a legitimate Pakistani spelling," says Dr. John Hermansen, chairman/co-founder/CEO of Language Analysis Systems. "That one difference allowed him to commit this crime unimpeded," he says, because no one thought to associate the name ‘Kasi' with ‘Kansi.' "It's a problem that can be stopped," Hermansen asserts.
And he ought to know. A Certified Knowledge Engineer (as designated by the International Association of Knowledge Engineers), Hermansen is a computational linguist with more than 20 years' experience helping the federal government decode and understand multicultural naming conventions by analyzing their linguistic and computational properties. After drafting the linguistics requirements for the U.S. State Department's visa processing system upgrade in 1984, Hermansen teamed with Dr. Leonard Shaefer to form Language Analysis Systems, a Herndon, Virginia-based company created to help the government "improve the quality of name-matching for both visa issuance and border security."
Over a 15-year period, LAS worked exclusively as a consultant to various government agencies, including the U.S. Departments of Defense, Justice, State, and Treasury; the Federal Bureau of Investigation; and the Immigration and Naturalization Service. By 2001, says Hermansen, more than 90 percent of LAS' revenues were generated from sole-source-justified government contracts. (Such contracts are awarded by the government to companies considered to be the single source of expertise in a given area.)
The September 11, 2001, terrorist attacks changed a lot of things for a lot of people; they also precipitated an evolution in LAS' mission statement. Following the attacks, FBI and INS agents used LAS' Name Reference Library software—which provided common cultural variations of the names listed on the flight manifests of the planes used in the attacks—to track several of the hijackers to their Florida connections. As the government moved to lessen its reliance on consulting services, LAS realized that it would have to become a product-driven company to stay solvent.
By 2002, LAS had launched a commercial division to bring the tools that "the men and women on the front lines at the border patrols and consulates were using" to businesses that wanted "to do a better job of understanding and communicating with their customers," says Hermansen. Today, the vast majority of LAS' revenues are generated through sales of its name- recognition and -analysis solutions for intelligence and security, homeland security, law enforcement, financial, and data marketing applications.
Drawing from a data archive of nearly one billion proper names used worldwide, LAS software uses patent-pending technology to identify names by culture and confidence factor (NameClassifier); search for multicultural names in a database (NameHunter); phonetically search names (MetaMatch); format names for data storage; generate all known variants of a name (NameVariationGenerator); generate equivalent names across cultures; generate additional attributes of names (NameGenderizer); and train field personnel in advanced multicultural name-searching techniques (Name Reference Library). (Commercial clients include Dun & Bradstreet and Lockheed Martin.) The company also licenses its solutions to OEM partners such as IBM, Microsoft Corporation, and Sun Microsystems, and provides certification to professional name-searchers in the government.