9 Comments

Summary:

Updated. Google’s translation services have been called all sorts of things over the years, from “incredible” to “woeful” and — by one particularly disgruntled user — “a total disgrace.” But here’s one to add to that pile: “unique.” At least, the U.S. Patent Office thinks so — enough to […]

Please maintain the cleansiness of the toilets

Updated. Please maintain the cleansiness of the toiletsGoogle’s translation services have been called all sorts of things over the years, from “incredible” to “woeful” and — by one particularly disgruntled user — “a total disgrace.” But here’s one to add to that pile: “unique.”

At least, the U.S. Patent Office thinks so — enough to award Google Translate a patent.

Update: This one The patent, issued in September, appears to have skipped under the media’s radar, but in and it was brought to my attention by a filing with the U.S. Patent Office published last week, which noted that the company was awarded the rights for what it calls “displaying original text in a user interface with translated text.”

That means it’s not the actual act of translation we’re talking about here, but the act of presenting the translation back to the user alongside the original text. In Google Translate’s website tool — which pops up when you visit a foreign-language site and currently supports more than 50 tongues including everything from English, Arabic and Chinese to Catalan and Welsh — the service translates the actual web page into your chosen language and displays the original text in a pop-up box when you hover over a piece of translated text.

Whether or not you think it’s right that it’s possible to get a patent for a method of presentation, rather than a process, it could be an important piece of news for several reasons. Translation is a pretty hot new area that’s advancing quickly. In the last year or so, we’ve mentioned services like Linguee and Babylon which could potentially be affected by Google’s ownership of this patent.

It’s increasingly obvious that translation is a big part of Google’s plans for the future, whether it’s via the Star Trek-like conversations mode — where two people can speak to each other in different languages in near real-time — or what Eric Schmidt called augmented humanity. Accurate automated translation is the sort of revolutionary service that makes Google incredibly powerful, potentially indispensable and is appealing to all kinds of business.

In fact, it’s got friends in some pretty high places. At the end of last year, the European Patent Office said it was working with Google Translate to provide translation services for applicants across the European Union. The idea was to try to remove some of the headache for applicants, who currently have to file a copy in English, French or German and in any of the languages of the European countries it wishes its patent to operate (that’s 37 states and, by my count, another 23 languages).

However, there’s one final note. The application, first filed in 2006, names a septet of Google engineers and designers, but there’s one notable omission: Franz Josef Och, the German computer scientist who runs Google Translate. He first designed the system after complaining that existing machine translation systems were inadequate (he even won a Darpa-sponsored contest for speed translation back in 2003), but his absence from the paperwork could be the strongest indication needed that this is less about the technology than the interface.

Image courtesy of Flickr user engineroomblog

Related content from GigaOM Pro (subscription req’d):

You’re subscribed! If you like, you can update your settings

  1. What does Google’s patent on Web based translation mean for WordLens?…

    Google got the patent for web-based translation (http://gigaom.com/2011/01/25/google-grabs-patent-for-web-based-translation/). Does this mean that WordLens needs to pay them a licensing fee?…

  2. JordanWalbesser Tuesday, January 25, 2011

    The original patent (US 7,801,721) was awarded on September 21st 2010. The filing linked in the article is to an application for a patent filed on September 20th, 2010, not a granted patent.

    A published application does not mean that the government is granting a patent, although it is one step in the process to get to that point.

  3. “At least, the U.S. Patent Office thinks so — enough to award Google Translate a patent,” “the company was awarded the rights.” Ummm no, Google hasn’t been awarded anything yet. All the patent office has done is publish the application that Google submitted. It will be some time before the application is examined and any rights, if any, are awarded.

  4. . . . but they were awarded a predecessor patent, 7,801,721, which is very similar to the continuation application referenced in your article.

  5. Bobbie Johnson Tuesday, January 25, 2011

    Hi Steve, a bad choice of words up above. When I said it “skipped under the radar” what I meant was that the original patent (awarded in September) had gone unreported: the continuation documents merely alerted me to the fact that the original patent had been granted. I’ll try and update the post accordingly to avoid further confusion.

  6. Will Google’s web translation patent hold up in court? See blog post I wrote on http://bit.ly/hyOZEZ

  7. Prevod Engleskog Monday, January 31, 2011

    Translation software uses algorithms and relies mainly on language usage statistics and word substitution. At its core, automated translation is an attempt to simplify human language and it very seldom, if ever, produces a viable and usable translation.

    Professional human translations, on the other hand, are created exclusively by professional certified human translators. They rely on extensive knowledge of both the source and target languages and the ability to naturally recognize and incorporate all the subtleties of human languages, taking into account such considerations as the cultural context, intended market, domain, target audience, etc.

    My vote goes to human translators, but I would never underestimate the power of hi-tech development.

    1. Prevod, I have to disagree with your statement that a computer “very seldom, if ever, produces a viable and usable translation.”

      It is true that Fully Automatic High Quality Machine Translation (or FAHQMT), is now impossible and this won’t change for a long time. But even today’s versions can be helpful today, to translation professionals and to our clients alike.

      First, humans sometimes make mistakes that a computer simply doesn’t make, such as spelling and numerical values. Second, sometimes there’s simply no time to wait for a human to do the job.

      Google uses a corpus of officially translated texts in many fields, especially those related to international organizations working in the fields of medicine, law, finance and commerce. In those fields, Google produces translations that are, as an acquaintance put it, “scarily adequate.”

      I have found that it frequently generates translations that may be totally unsuitable to send to the customer, but which nonetheless can serve as a starting point for refinement.

      Some of my clients use Google to get a rapid sense of what a document or a web site means, and whether or not it’s worth sending it for a complete translation.

      I endorse your argument that human beings will remain the standard for high-quality translation, but it’s incorrect that machine translation is not viable or useful.

  8. Interesting! Google Translate’s functionality, though, still leaves a lot to be desired.

    We just wrote an article about some of the more “interesting” patents that have come through our office recently …

    http://www.schreibertransblog.com/sti/2011/02/patents-can-be-fun-%E2%80%93-or-at-least-interesting-%E2%80%A6-really/

Comments have been disabled for this post