They claim to have found a real breakthrough at 'contextual understanding' of Text, and thereby finding new powerful ways for machine translation...

But this is all behind a 'patent-pending' wall of buzzword-ness, I'm searching for some pointer of what they think they have found that is so revolutionary...

Washington Post article onMeanigful Machines [via Maciej, THX!]

Abir's approach involves a variation of the second method. His company spent last year encoding his ideas into software algorithms that perform novel forms of pattern analysis that rely on phrases -- rather than words -- as the core unit of meaning.

Abir's system analyzes huge amounts of previously translated text -- such as United Nations documents -- and breaks matching sentences from different languages into paired fragments or phrases, storing them in a database. It also collects information about words that frequently turn up on either side of those phrases, in "overlapping" sentence fragments.

Aha, hmmm, I think I can see what they are going at... to me that sounds a little like Markov-Chains and some 'Compression Analysis' and then some... I guess the 'huge amounts of previously translated text' part is the crucial thing to make it work well.

Like Maciej I'm very curious to see it work... And I'm also very sad that it is going under the patent-hood of commercial big irons...

[ by Martin>] [permalink] [similar entries]

similar entries (vs):

• Idle Words at it again (# 8%)
• Happy Birthdays (# 8%)

similar entries (cg):

• :: phpPatterns() - Unit Testing in PHP (# 9%)
• Happy Birthdays (# 8%)
• Whatis AudioBooksForFree.Com? (# 6%)
• ResearchBuzz -- Another Undocumented Google Syntax -- Date-Based Searching (And some API Fun) (# 5%)
• can I post as you? (# 5%)
• wxLua 2.0 availeble (# 5%)
• on how (# 5%)

Martin Spernau
© 1994-2003

Big things to come (TM) 30th Dez 2002