Like vacations on the moon, jetpacks, and other Jetsons-like inventions, machine translation of human languages has been a long-predicted and much-delayed technological advance.
, a Pittsburgh start-up that just won Phase One SBIR funding, aims to apply its discoveries to make machine translation practical as well as grammatical.
“The fundamental problem is that human languages are highly ambiguous. We’ve been researching for 40 years, and success has been just over the horizon. Now, the approach is different, thanks to the Internet and the availability of large amounts of data,” says company co-founder Alon Lavie.
In his day job as a professor in Carnegie Mellon University's Language Technology Institute,
Lavie develops fundamental algorithms that are not specific to language pairs. With Safaba--a meld of the Hebrew words for inside and language--he’s looking to develop core software for commercial users.
“With Safaba, we’re working on more concrete issues. We’re making the state of the art more usable in the translation industry,” says Lavie. He’s partnered with Bob Olszewski, a colleague in the School of Computer Science at CMU, on the start-up housed at CMU’s incubation center.
Incorporated in June 2009, Safaba has raised “only a small amount” of funding,” says Lavie. The SBIR grant will allow the firm to develop its software and create a plan for full-scale commercialization.Source: Alon Lavie, SafabaWriter: Chris O’Toole