Arab names put spell on US hunt for terrorists

The Telegraph (U.K.) ^ | 07/02/2002 | David Rennie

[Pokey78]

America's intelligence agencies are struggling with a major hurdle in the war on terrorism, it was reported yesterday: they cannot agree how to spell Arab names.

They are having trouble tracking terrorist suspects, as non-Arabic speaking agents convert Arabic names into English, using any one of dozens of competing spelling systems.

One CIA agent reported finding more than 60 versions of the name of Libya's ruler, Muammar Gaddafi, on one computer database.

Ageing databases have been thrown by the inability to search different spellings of the same name, such as Mohamed or Muhammed.

The FBI director, Robert Mueller, last month admitted to the Senate judiciary committee that his agency's computers would be unable to find his own name, if it was slightly misspelt.

In a separate problem, intelligence officials have been led to entirely innocent Arabs who happen to share common names, such as Muhammed, Sheik, Atef and Atta.

"I can't tell you how many Mohammed Attas we've run across," one official told USA Today, referring to the leading September 11 hijacker.

[Free the USA]

Bump

[RightWhale]

That's right, there is no standard English spelling of Arabic words or names. Even phonetic systems vary. The only solution, if this current problem is going to be with us for some time, is to learn Arabic. It can be done.

[Maceman]

The only solution, if this current problem is going to be with us for some time, is to learn Arabic. It can be done.

I'm not a software guy, but how hard can it be to plug in the Arabic characters, and generate all the potential English phonetic spellings? Then a database query can be constructed that includes all of them.

What am I missing? Just because these terrorists' names sound like the noise you make when you have a hair in your throat, is no reason we can't track 'em.

[RightWhale]

like the noise you make when you have a hair in your throat

That's true, they do. Even just a little knowledge of Arabic makes dealing with the phonetic systems a lot easier. A course in 'Arabic for one-time tourists' tape would do wonders.

[eno_]

Ages ago AT&T invented a string comparison algorithm that effectively matches homophonic names by discarding letters. Crude but effective, and widely used in directory assistence.

[Paleo Conservative]

Ages ago AT&T invented a string comparison algorithm that effectively matches homophonic names by discarding letters. Crude but effective, and widely used in directory assistence.

But you can't expect the CIA to use software they didn't write themselves.

[gaspar]

Ostensibly, the US Government is required to use the orthography used by the Library of Congress, and place names are to follow those used by the US Board on Geographic Names. That being said, most writers are not familiar with the systems. Researchers familiar with Arabic have dropped the letter "e" from the vowell list except in the case of a dipthong. Unfortunately, the press for years had used an e in both Bin Ladin and in Muhammad. As for Qaddafi, the British usually spell it Gaddafy or a variant, and the French have their Kaddafi and variations. I feel for the researchers but the problem certainly is not intractable.

[Paleo Conservative]

I have no doubt that the problem is not intractable. I just wish that the would government use look to the private sector to help solve the problem.