“When I studied linguistics, the vast majority of the researchers were sure there would never be a real machine translation,” Oded Broshi, CEO of Alfabetic.Even though the Internet has democratized the way people communicate, receive news, and product information, it remains Western-focused and ignores the 70 percent of the world that can’t read English – about 1 billion people who are currently online.
If you are a content owner of a news website or blog monetizing your content through ads, millions of hits will never reach your site simply because potential readers can’t read it. A new Israeli company Alfabetic, is taking a major leap towards changing the way the non-English world consumes content, creating new business opportunities for content owners, while making the Internet a truly democratic forum.
The company was founded by linguist Oded Broshi, Alfabetic’s CEO, who has a background in brain research. “When I studied linguistics, the vast majority of the researchers were sure there would never be a real machine translation,” he tells ISRAEL21c.
A.I. and real human translators
“My vision came from Hollywood – to conquer the world. Everyone wants to see movies from Hollywood. Subtitles and dubbing made it possible, and most of Hollywood’s revenue now comes from foreign sales. We thought we could do the same for the hottest foreign content in English,” says Broshi, whose company has built a translation machine based on artificial intelligence. “There is a lot of hunger and curiosity out there,” he adds.
Typical English sites that Alfabetic could translate and syndicate, would include the technology news site Mashable, the celebrity gossip site TMZ, and the popular Autoblog, on everything car-related. They have already built a pilot site for the popular technology site TechCrunch, translating it to Spanish, and are now negotiating with the company.
Broshi explains that a new breakthrough in the academic world in statistical machine translations, has enabled superior machine learning of languages recently. These algorithms from the academic world paired with Alfabetic’s own technology, and an international team of translators, creates top-notch translated content, as though it were produced by the world’s best translators. But without the high costs, of course.
“The more put into the translation machine the better it gets,” notes Broshi. He offers an attractive business model – a three way split of ad revenues – for the content owner, the syndicate and Alfabetic.
Start with an RSS feed
Getting started is easy, says Broshi. The Alfabetic team first creates a clone of a news site. It then takes RSS news feed articles and plugs them into the translation machine, with no cost to the content owner. The output, which is produced in seconds, is then fed to a local team of translators – Chinese, Arabic or French for example – who tweak the translated material to ensure no machine-generated or cultural errors.
On top of that, a head editor in the foreign language reviews the final translation before it gets published, meaning newspaper editors at the New York Times will not have to worry about the quality standards of the material being translated.
“We’ve created a full ecology where we get an RSS feed from a major blog, we translate it with our servers near real-time, from there it goes to teams of proofreaders set up around the world,” says Broshi.
Broshi’s partner in the business is Arik Kopelman, a high-tech veteran who among other things co-founded Relegence, the financial news and information search technology company sold to AOL for millions.
Based in Kibbutz Tsorah, Alfabetic was founded in 2007 and now employs 10 people, plus an extensive number of outsourced translators around the world. The company is fully operational and is now looking for large content partners from the Western world.