Eli Abir, a former restaurateur and used car salesman, is hardly the person you'd expect to devise a breakthrough technology for machine translation of languages, a problem that has proven tough to conquer.
But after patenting a system that allows non-English speakers to surf the Internet in various languages, Mr. Abir felt ready to build a computer brain for translating languages. "The concept is that language has a DNA, and when you isolate the DNA of ideas, it makes no difference whatsoever what language you express it in," says Mr. Abir, founder of New York-based Fluent Machines.
Investors in the original project scoffed at the newer proposal at first, but they have since stopped laughing. Apple Core Holdings, a private-equity group in New York and an investor in the first project, has doled out $2.1 million more. The "example-based" translation system uses artificial intelligence to match sentence fragments in corresponding languages, then reconstructs the pieces like a jigsaw puzzle to create a coherent translation.
Jaime Carbonell, a leading machine translation expert from Carnegie Mellon University, has become a fan--and a board member. He says Mr. Abir has "some kind of intuitive genius feel for how language works" and that the concept is the biggest breakthrough in the field in 20 years. It will take time to see if it translates into a viable commercial product. But Allied Business Intelligence, a market research group, estimates that computer translations accounted for only $73 million of a $5.7 billion translation market in 2001. That potential sounds sweet in any language.
Write to Avi Machlis.
|ABOUT US TERMS OF SERVICE ETHICS POLICY MEDIA KIT HELP|
|Copyright 2002 Red Herring Magazine|