With the motto “The Language of Business. The business of Language”, Gala will hold its 10th annual congress, which will take place in Boston (US)...
2 min read
12/10/2017
Pangeanic will introduce its business technologies for translation companies at the Networking Days of the European Language Association in Bucharest, 12-13th October 2017. Elia's Networking Days are leading get-together for international translation companies in Europe doing business on the continent, vendors of translation services to public administrations. Manuel Herranz will address an audience eager to find out more on neural machine translation, how affordable it is to own a massive language database based on Elastic Search ( ActivaTM), fully customized for translation professionals and of course Cor, an easy-to-install web-based ordering system. All these technologies can make any translation agency, large or small, more competitive and improve the speed of any translation cycle for your clients. How do you see your business in the next 5 years? That is the key question Manuel will be directing to the audience. Do you think you will still manage jobs via email? How many translation memories can you organize? Can you leverage from all your work (or most of your work) to profit from existing TMs? Is neural translation the real break in machine translation we have all been waiting for? Centralized TMs for massive leveraging were until now the realm of powerful translation companies that could afford to invest heavily in technology and servers. ActivaTM is a result of an EU research program, which makes it very affordable. All your translation assets can be placed in a single, stable database. Due to its Elastic Search origins, ActivaTM is ultrafast. It can reverse language pairs and create new language combinations via its "triangulation" feature. It also create translations when the new sentence has only changed 1 or 2 words by substituting numbers, emails, even a word. Lastly, ActivaTM links to neural machine translation - the engines we create for your or general online engines. 
Cor is probably one of our best-liked business technologies for translation companies. It is a web-based online document translation service with a web crawler added to it. It can help you track your client's new publications (or deleted parts), count the number of words on a website, convert it to xliff for translation and send it back for publication. It is not proxy translation. It installs as a frame in any website, so as long as you have a database of translators and a pricelist, you are ready to go! Cor also comes with an online vendor tool and quotation tool so any translation company can provide website translation services anywhere in the world. Lastly, our neural machine translation developments show the starting point of the technology is three or four times higher than statistical in terms of human evaluation. Having walked the statistical walk for the last 10 years, we are pretty optimistic about it. We will highlight the current limitations and heavier hardware requirements which make neural the thing to use but not the thing to host.
Cor is probably one of our best-liked business technologies for translation companies. It is a web-based online document translation service with a web crawler added to it. It can help you track your client's new publications (or deleted parts), count the number of words on a website, convert it to xliff for translation and send it back for publication. It is not proxy translation. It installs as a frame in any website, so as long as you have a database of translators and a pricelist, you are ready to go! Cor also comes with an online vendor tool and quotation tool so any translation company can provide website translation services anywhere in the world. Lastly, our neural machine translation developments show the starting point of the technology is three or four times higher than statistical in terms of human evaluation. Having walked the statistical walk for the last 10 years, we are pretty optimistic about it. We will highlight the current limitations and heavier hardware requirements which make neural the thing to use but not the thing to host.