Process Innovation Challenge (PIC) #7 takes place in Silicon Valley at LocWorld 41 – 7th November 2019 at 1.30pm. As always, innovators will pitch their innovative ideas to a panel and the audience. This year there is one round only, so the pitches needs to be PICture perfect and completed in 4 minutes. This years dragons are Neftalí Jovel (Indeed), Alessandra Binazzi (Asics), Konstantin Savenkov (Intento). The innovators and innovations are :
How to Get the Most Accurate MT Translations across All Industries
Heather Shoemaker (Language I/O)
Language I/O has created a multiple machine translation (MT) engine approach to user generated content translation such as customer support chats and tickets. Our SaaS solution gathers a variety of human quality measurement/distance inputs combined with complexity data to select the best MT engine for chat or ticket translation.
Automation in Regulated Environments — The Final Frontier
John Tinsley (Iconic Translation Machines)
Language is not an excuse when it comes to the law. In ediscovery, lawyers need to understand the data regardless of the language, but it’s complicated. They’re often sitting on terabytes of data in different formats and languages, and all they know is they need it in a language they understand in their searchable database, and they needed it yesterday. We have built an integrated neural machine translation solution for ediscovery that has been adapted to the vagaries of global litigation and processes vast amounts of data on-the-fly, at scale and securely.
Lucio Gutierrez (Intuit)
Intuit builds financial products that must be well localized and translated in order to gain and keep customers’ trust. To achieve this, we have created a tool called MagicUI. This tool enables content editors to localize and translate text within the user interface (UI) and push it live within minutes.
A New Way to Setup Source and Target Languages
Robert Rogge (Zingword)
By altering the traditional data model for language pairs and making source and target languages distinct features of documents, we have achieved the most complex language setup possible in the simplest documents/languages user interface ever created in the localization industry. Let’s get the basics right and go from there.
Lori Silverstein (SPi Global)
We present dubbing based on AI processing of synthetic voices and the latest advancements in voice morphology applied to text-to-speech. While not intended to replace human dubbing, the solution supports additional options for users to engage with content in their own language and increases discoverability for content otherwise not available to certain target audiences.
Automating Linguist Selection Using Topic Modeling
Dalibor Frívaldský (Memsource)
An automated linguist recommendation system based on unsupervised topic modeling (content understanding). For any new content, the system can automatically find the most appropriate linguists to assign to the translation task.