Software Developers and LSPs Enhance Machine Translation Capabilities

Article Details

Global Watchtower

Common Sense Advisory Blogs

Posted by Donald A. DePalma on February 14, 2013 in the following blogs: Technology, Translation and Localization
Pages \| \|

In recent months many developers of machine translation (MT) technology and supporting tools have told us about new features that will increase the quality and usability of their products. New to the ranks are language service providers (LSPs) that are actively investing in production models employing MT. Here are some noteworthy additions, listed alphabetically, that have come to our attention:

Asia Online (Singapore) has just opened its Language Studio 3.0 to select customers for beta testing, and will release it to all customers in next few weeks. CEO Dion Wiggins said that the new version includes advanced features for customizing engines, a new management interface, and new runtime control features.

euroscript (Luxembourg) told us that it strengthened its MT development team and technology to reflect its work on the TaraXÜ machine translation project. CEO Mark Evenepoel said that the company will soon relaunch its MT services with language quality enhancements for French- and German-based language pairs, better workflow integration, and support for more source formats. Meanwhile, the TaraXÜ project continues work on language quality assessment, rules for automatic pre-evaluation of MT quality, and automating the terminology feedback loop.

KantanMT (Ireland) released version 7 of its Beta program, adding an analysis tools to predict the human effort required for post-editing and a dashboard for monitoring projects. The company also added support for OpenOffice file formats.

Lexcelera (France) launched its LexWorks subsidiary in Canada to offer end clients MT solutions, ranging from customizing rule-based, statistical, and hybrid engines, to deployment and integration, to post-editing for publication quality. Separately the company’s founder, Lori Thicke, announced that John Papaioannou, a localization veteran and formerly a buyer of MT and other services, will join the company on March 1st as its CEO.

Lingo24 (UK) recently hired a team of MT specialists led by Andy Way to integrate MT into its Coach translation environment and to offer machine translation services to end clients as well as to freelancers, single-language vendors, and multi-language vendors without their own MT.

Ortsbo (Canada) announced a partnership in China to bring its MT cross-language communication services to that market. Its Chengdu operation will be in the middle of the Sichuan National Promotion Base for the Game and Animation Industries. The company’s 2012 summary reiterated its focus on addressing global customer care and market communications, gamification, and specialized social networks.

PangeaMT (Spain) has been a pioneer in do-it-yourself (DIY) deployment of Moses MT technology, particularly in helping LSPs apply their knowledge and experience to training their own engines and in serving commercial and government end clients that want to manage and grow their own MT ecosystem. Last year the company launched PangeaMT SaaS Power for customizing, hosting, and managing MT engines.

Safaba (US) ratcheted up its marketing effort, focusing on the enterprise MT product that it launched last year. Founder Alon Lavie told us that the company is now preparing a solution that addresses persistent post-editing challenges such as the translator’s ability to control and affect MT output quality.

SDL announced machine translation enhancements to its Language Platform and BeGlobal MT technology. It also coined the term “big language.”

SYSTRAN (France) announced SYSTRANLinks, a proxy server that automatically reproduces a company’s website in languages specified by the site owner, machine-translates content on-demand, caches the MT output for post-editing if desired, and provides a platform for collaborative, workflow-based editing of the site.

TAUS (Netherlands) announced a toolkit for Moses developers that includes new features such as applications for training, testing, scoring, and aligning text. It also packages a sample model for experimentation, and now supports a broader range of operating systems, including Windows, Mac, and several flavors of Unix. In late 2012 TAUS also announced its Dynamic Quality Framework, a way to select the quality assessment model best-suited to an application, a knowledgebase of best practices for evaluating results, and shared tools for benchmarking.

Welocalize (US) has been enhancing its weMT program for evaluating the effort required for post-editing to build more accuracy into pricing and productivity discussions with its clients and the linguist supply chain. Vice president of language tools Olga Beregovaya told us that the company has also extended the OmegaT-based post-editing evaluation platform that it is developing in collaboration with CNGL in Ireland, and that it’s also working with the OmegaT community to integrate OmegaT with its GlobalSight translation management system.

These announcements and briefings point to the fact that MT is quickly gaining momentum among developers, language service providers, and, as our preliminary research indicates, among buyers of translation services, too. While by no mean exhaustive, this list demonstrates the strong investment and innovation driving the machine translation sector.

We just started a new research study on how enterprise (buy-side) organizations use machine translation. If you buy post-edited machine translation or MT software, please tell us about your experience. Even if you don’t use MT, our survey has some questions for you, too. And if you’re an LSP that uses MT or a software vendor that develops it, please brief us on your services or products.

Categories