VALENCIA, Spain--(BUSINESS WIRE)--PangeaMT, Pangeanic’s Translation Automation Technology Division, is now offering language service providers, corporations, institutions and other user segments PangeaMT DIY SMT, a unique opportunity to deploy customized, in-domain statistical machine translation (SMT) engines that they can build or update as need be. Owners of a PangeaMT DIY solution will enjoy maximum flexibility, technology independence, and cost-effectiveness as there is no need to contact the MT provider to generate further engines or update or retrain existing engines any more! Furthermore, they will be able to integrate glossaries and DNT features to gear machine-translation output towards even more predictable results.
The solution comes with an EN (English) to/from FIGS (French, Italian, German, Spanish) training structure, which is expanded at will according to domains to reflect the corresponding engines that users will need to generate and update in those language pairs. Engine creation scripts and language models for other language pairs can be built upon request and incorporated in the DIY infrastructure. In other words, after having first-time modeled engines, users can go solo, not only for making unlimited number of MT requests through an intuitive web panel interface, but also to generate more engines by picking and adding data pertinent to other domains or post-edited material, fixing terminology via a glossary feature when required. This is all paired with a control panel that caters for user creation and management, assigning engine(s) to user(s) or company department(s). The panel displays valuable statistics about engine training status, vocabulary and number of words, BLEU quality metric percentage, etc.
It is the first time that an MT provider decides to empower their clients so much, providing them the possibility to grow and customize their own solution according to client and market needs with all the tools, scripts, web interfaces and base engines. The objective is to enable users to create translation engines and experiment without having to contact the MT provider and without having to release internal data. This is the maximum expression of PangeaMT’s mission, that is, to bring democracy to the machine translation arena and create a platform to make the technology affordable and usable for as many users as possible.
But lowering the difficulty in deploying, experimenting, and expanding your own MT solution are not the only advantages that PangeaMT DIY SMT brings along. "In this day and age machine translation technologies from our competitors are mostly offered under a license system with MT corporate clients locked in for a long time, after having purchased a pretty expensive software annual license. Alternatively, there is the per-word SaaS option with heavy set-up or trial costs – neither of which is often perceived as cost-cutting in the customer organizations. This prevents many prospective buyers from believing that MT is a language technology worth investing it providing real, fast ROI and faster speed-to-market!”, regrets Elia Yuste, Pangeanic’s Business Development Manager and PangeaMT’s Scientific and Commercial Lead. With the DIY SMT solution, the PangeaMT team wants to produce an MT offering that converts MT skeptics into pro-active, determined MT players, not binding them to use X or Y software or pay for a license renewal the following year.
This intrinsic operational freedom and independence of the PangeaMT DIY solution is also the best guarantee for data confidentiality of your translation assets. PangeaMT, being a technology spin-off of a 15-year old translation company, had to be a solution that guaranteed the critical principle of confidentiality. “We have always been totally against data sucking, overtly or secretly as a means of improving MT vendors' systems,” declares Manuel Herranz, Pangeanic’s CEO and PangeaMT’s main evangelizer. “The engines we have built for our clients have always benefited from the necessary data analysis and consultancy that every MT customized development requires. We found a common problem, however, in that providing data in the form of TMs is a stumbling block for many potential machine-translation adopters who can use their data sets for training but are weary of making them available even under NDA. Now with the DIY solution, LSPs, translation departments and institutions can make use of their own private data for training, resting fully assured that they count upon the technical means to train and retrain their own engines without any intervention from us or any other MT provider,” Manuel stresses.
The PangeaMT team will continue to build engines for those users that feel they do not need DIY capabilities. Other users may opt to have engines created in one or two domains and language pairs, familiarize themselves with MT and post-editing multilingual content production workflows, and then go on to adopt a full DIY solution. We anticipate that this kind of intermezzo period may be preferred by some and will also serve them gladly. “However, we are dreaming about the day where all our clients will go for DIY solutions,” Elia reckons. “This really is the way forward from a technological perspective, if we really want to make sure that Jaap van der Meer's prediction “Let a thousand MT systems bloom” at the TAUS conferences materializes. Perhaps users will start with small scope engines, but once they experiment the power of data-driven customized engines, they will soon apply multilingual communication solutions that are more agile and less expensive.”
To conclude, Manuel Herranz notes that “just as large OS software companies do not charge you for the number of documents you create after buying their famous word processing packages (and even giving you some basic text editors for free), we do not understand that engine creation, experimentation and automation should be penalized.” The PangeaMT team wishes users to become as independent as possible so that they can focus on their own business automation, research and address their own needs and innovate even further. We want to continue to contribute to the multilingual translation automation industry with MT democratizing solutions.
Feel free to contact us for more information about PangeaMT engines and the DIY SMT solution.
About Pangeanic and PangeaMT
Pangeanic, the 15-year old, Spanish leading provider of translation, publication, software and web localization and manual-making services, works independently and in association with sister offices in Japan and China for (cross-national) institutions, famous multinationals, government agencies and other language companies all over the world. Our built-in quality service philosophy is accompanied by the latest computer applications and state-of-the-art language quality assurance (QA) technology. PangeaMT is the company’s Machine Translation (MT) R+D+I division and a European leader in the implementation of independent, open source statistical and hybrid translation automation solutions with a record of customized developments. PangeaMT brings together a team of professionals with international expertise in computational linguistics, translation, content and knowledge management. PangeaMT’s real-world implementations and technology advances have been presented in well-known industry events and discussion fora.