Lingotek Brings Translation To Social (And Vice Versa)

The system uses crowdsourcing to provide translations, and integrates with Jive, Drupal, and Oracle Content Management to handle social content.
10 Crowdsourcing Success Stories
Slideshow: 10 Crowdsourcing Success Stories
(click image for larger view and for slideshow)
If you are building a corporate social network for employees around the world, are you speaking their language? Or is something being lost in translation?

The danger, as Lingotek CEO Rob Vandenberg sees it, is that you wind up with "a worldwide implementation of a social business software solution, and now you've enabled everybody to have siloed communication--where , China can talk with China, but they can't talk with us."

When Lingotek signed on as one of the launch partners for the Jive Apps Market introduced with Jive 5.0, part of the logic of the deal was that at least some of the content generated on social networks needs to be intelligible to an international audience. This is one of a series of social and Web content management deals the company has struck in the last few months, including integrations with Oracle Content Management and Drupal, (via Acquia, which provides commercial support for the open source Web content platform).

Lingotek also embraces the social media concept of crowdsourcing, allowing online communities to collaboratively translate content on a volunteer basis, or in exchange for bragging rights on a leaderboard for contributors. Community translation can be a sort of in-between step between automated systems and traditional professional translation. Lingotek's Collaborative Translation Platform provides a workflow for managing translations, either independently as or a component embedded in a content or social software platform.

Prominent customers include Adobe, which uses the Lingotek platform to translate content for its Adobe Groups support forums. Lingotek also does business with federal agencies, and some of its funding comes from In-Q-Tel, a non-profit associated with the CIA that invests in technologies for intelligence gathering.

The traditional model for translating content, which relied on agencies farming out work to professional translators, can't keep up with the pace at which content is generated by social media, Vandenberg said in an interview. At about 23 cents a word, professional translation also is too expensive to apply to the masses of social media content, he said. Lingotek's answer is to acknowledge the range of different types of content that require different treatment.

The least expensive is machine translation--an instant approximation generated by software that has a statistical understanding of how words and phrases in different languages are related to each other. Content can be run through public services from Google or Microsoft's Bing, through machine translation products like Language Weaver, (recently acquired by SDL), or through custom translation engines trained for specific subject matter. For self-service purposes, machine translation is often good enough to give the reader the gist of a page or post.

Where that's not adequate, community translation (the crowdsourcing angle) lets people work together to improve a translation to the point where it's usable for their purposes. Lingotek helps that process along with collaboration tools and translation memory libraries of previously translated content.

Finally, professional translation services can be applied to content that must project a professional image. Lingotek can procure those services on behalf of its clients, although many of them use the platform to manage work in cooperation with other translation services firms, Vandenberg said.

Even where professional translators are part of the mix, Lingotek recommends holding down the cost by letting content work its way up this food chain. Machine translation is usually, but not always, the first step providing "a jump start on translation," even though it's not perfect, Vandenberg said. "Most people say that if it's not 60% to 70% accurate, then it's not worth it to edit--they should just rewrite it," he said, but usually it's better than that.

"An employee might say, here's an interesting relevant blog post, let me get this into my own language so I can share this with people in France," Vandenberg said, so in that case, community review--sharing the document with a few other people who are bilingual employees who can help improve it--might be fine for internal sharing. "Or, if this is going to be released publicly as a white paper, maybe the machine translation hits it first, and then the community post-edits it, and a professional translator reviews it."

Vandenberg said Lingotek hopes to support more social and content management platforms and has been working with Microsoft partners on SharePoint integration. "We'd love to get deals with everyone, but Oracle, Jive, and Drupal will keep us busy for a while," he said.

You can't afford to keep operating without redundancy for critical systems--but business units must prioritize before IT begins implementation. Also in the new, all-digital InformationWeek SMB supplement: Avoid the direct-attached storage trap. Download it now. (Free registration required.)

Editor's Choice
Brandon Taylor, Digital Editorial Program Manager
Jessica Davis, Senior Editor
Terry White, Associate Chief Analyst, Omdia
Richard Pallardy, Freelance Writer
Cynthia Harvey, Freelance Journalist, InformationWeek
Pam Baker, Contributing Writer