Slashdot Mirror

← Back to Stories (view on slashdot.org)

Romancing The Rosetta Stone

Posted by Hemos on from the cool-story dept.

Roland Piquepaille writes "Not only this news release from the University of Southern California has a fantastic title, it also has a great content. This story is about one of their scientists, Franz Josef Och, whose software ranks very high among translation systems. "Give me enough parallel data, and you can have a translation system for any two languages in a matter of hours," said Dr. Och, paraphrasing Archimedes. His approach relies on two concepts, gathering huge amounts of data, and applying statistical models to this data. It completely ignores grammar rules and dictionaries. "Och's method uses matched bilingual texts, the computer-encoded equivalents of the famous Rosetta Stone inscriptions. Or, rather, gigabytes and gigabytes of Rosetta Stones." Read my summary for more details."

4 of 486 comments (clear)

  1. Re:Where can I download his software? by Anonymous Coward · · Score: 0, Redundant

    Not only this launching of the news of the university of southern California has a fantastic title, he also has a great content. This history is near one of its scientists, Franz Jose Och, who software aligns very above between systems of the translation. "déme enough parallel data, and you can have a system of the translation for any two languages in a question of hours," said to Dr Och, paraphrasing Archimedes. Its approach trusts two concepts, compiling enormous amounts of data, and applying statistical models to these data. It totally does not pay attention of rules and dictionaries of the grammar. the "method of Och uses the matched bilingual texts, the equivalent ones computer-codified of stone the famous inscriptions of Rosetta. Or, something, gigabytes and gigabytes of stones of Rosetta." Read my summary for more details."

  2. Re:The vodka is strong but the meat is rotten by aastanna · · Score: 0, Redundant

    I thought the whole point of this algorithm was that if you give it enough data, eventually it would see that phrase and remember it, giving a perfect translation.

    The only problem with ignoring grammer and syntax and just getting massive amounts of data would likely be storage space and training/translation time.

  3. Re:DARPA by zebs · · Score: 0, Redundant

    Kneejerk /. forth-order: OOOOOooowwwwww me knee. I'm getting to old for this :(

  4. Re:Oh god... by Dread_ed · · Score: 0, Redundant

    )

    My EYES!!! DEAR GOD MY EYES!!!

    yes mr. filter software I am yelling, you would too if someone poked out your eyes.

    --
    When the only tool you have is a claw hammer every problem starts to look like the back of someone's skull.