Introduction
This post contains some of the more technological projects I’ve worked on for the past few months. For your convenience, I’ve included some assets from these projects to help contextualize what I’ll talk about briefly.
Part One: Machine Translation Project
Quick File Access
Pilot Proposal: [Download] (PDF)
Lessons Learned: [Download] (PDF)
I was fortunate enough to have an opportunity to train a machine translation engine from scratch along with some of my colleagues! We chose to try and train an engine capable of handling Dungeons and Dragons campaigns and supplemental materials; I’ve included our project proposal and lessons learned here.
Part Two: Translation Memory Systems; An Applied Look
Quick File Access
Presentation: [Download] (PDF)
In the past few weeks, my colleagues, Chris Dean and Marianna Guedez, and I were fortunate enough to get a demo of Keywords Studios’s XLOC. Many thanks to Rebecca Guttentag and Jenny McKearney with Keywords Studios, who went above and beyond to give us a look at their Content Management System. They were incredibly accommodating and were enthusiastic when answering our questions, without their help this review would not have been possible! We’ve included our video presentation here, but you can find it below as well.
We went over it in our presentation, but XLOC is more like a CMS than a TMS. Rebecca described it as something in between, but Jenny told us that it leans more towards managing content than translations, the reason for that being that the developers of XLOC did not want to have to compete with big names in TMS technology. That is unfortunate because I think XLOC would greatly benefit from some TMS-like capabilities; that being said, the fact that XLOC integrates into memoQ and that they have a Memsource integration in development is a great asset to them and their commitment to an on-the-cloud experience. Another thing they would greatly benefit from, although this is still along the lines of a TMS, is in-context preview capabilities. This would be incredibly useful in seeing how string expansion would affect a game, whether the text truncates or causes other unforeseen issues. For that to be possible, it would take some changes to the way video game localization is done, which is unlikely without some work.
All that being said, I’d like to reiterate a lot of the things that XLOC does well. We talked about it in our presentation above, but it deserves a second mention. XLOC’s ability to handle any file type, even proprietary file types, is among one of its greatest strengths. Jenny was confident when explaining how it works, stating that as long as the client provides the unique Unix string ID, string location, and where the client wants the translated content to be stored, XLOC is fully capable of handling any type of file. On top of that, XLOC is able to integrate with any technology as long as there is an API key.