Haiyi Lin

INTRODUCTION

This portfolio includes the NMT Training Project and the TMS Project.

NMT Training Project

This pilot project aims to estimate the work involved in training a neural machine translation (NMT) engine for military news from English to Simplified Chinese.

After the training, our BLEU score rise by 4.26, which is a promising result.

Proposal

Updated Proposal

Lesson Learned

Deliverables

Here is the link to the Deliverables of our project. Please feel free to download.

TMX Project

For TMX Project, we selected American Red Cross as our client. And we compared two TMX platforms for ARC, which are SmartCat and XTM.

American Red Cross is a nonprofit organization that aims to help all in need regardless of language through its strong network of volunteers, donors, and partners.

For our stakeholders, we have four sets. The first one is our project manager Karen. She’s the director of multicultural communications. The second one is our marketing team. We have several full-time marketing employees. All of them are bilingual and help on the site. The third one is our linguist. They are Spanish translator volunteers and TI vendors. The last one is the MT consultant, basically is Adam Wooden, and student volunteers.

So we will analyze two TMX platforms, SmartCat and XTM. SmartCat is a cloud-based translation and localization platform that connects businesses, translators, and translation agencies in a single connected translation delivery loop. For XTM, XTM Cloud is a translation management system with built-in translation, memory, terminology, and CAT tools.

This is a scoreboard for seven key business requirements. We give them weight points to prioritize the seven requirements, and we have a score for both XTM and SmartCat.

For PEMT workflow, we think it is relatively important because our project requires training our MT, so we think that our workflow is basically PEMT. That’s why it’s important. For XTM, it supports most MT integration and custom MT. However, we need to purchase MT APIs instead. And it only supports SYSTRAN in the Free version. However, SmartCat supports custom MT, and different MTs are included in the free plans with limited words. After comparison, we think that SmartCat is more suitable for PEMT workflow. That’s why we give SmartCat 9 points and XTM 7 points.

For our Cloud-based functionalities, both platforms are cloud-based platforms. So we gave the full 10 points for both platforms.

In terms of Volunteer-friendly, both platforms have good instructions. However, SmartCat is definitely simpler and prettier with UI and intuitive creatures.

By Tools and assets, we refer to TM and TB. XTM outperforms SmartCat because SmartCat has this unique feature of smart words with charges for TM.

In terms of Audio-visual content, both platforms support in-context reviews of videos. And we think the American Red Cross might need videos for marketing. But XTM only has a beta version and is not mature enough. However, SmartCat has a relatively mature subtitling tool. So it has two more points than XTM.

And in terms of Customizable project workflow, both platforms have workflow templates and support customization for workflow. So they achieved full points for this requirement.

And the last part is Cost-effectiveness. It is relatively complicated. And after some complex calculations, we figured that SmartCat charges around $1,500 more than XTM per year, which means that XTM may be more cost-effective. However, for a more accurate calculation and conclusion, we need to know the specific translation workload and numbers of users of the American Red Cross.

In total, XTM has 229 points and SmartCat achieved 242 points. Therefore, in conclusion, we think SmartCat would be a better option for American Red Cross as their TMX platform.

Please click here to view the presentation video.