Slator Buy-Side Report 2018 Actionable Insights From the Language Industry Buy-Side
Features 23 buyer profiles along industry verticals.
Once the Task Force had pooled their combined knowledge and experience into a treasure trove of TM tips and tricks, the initial draft was supplemented by input from 32 members of the GILT Community (18 buyer organizations, eight LSPs and six machine technology (MT) technology providers) to create a “living” document of TM best practices.
The “How-to” of TM Management
Through its self-serve best practice guide, the TM Task Force hopes to share the benefits of their own research and experience with the localization buyer community, and specifically aims to help those who manage TMs to augment their existing practices and increase their knowledge of “translation memory management, as it relates to translation management systems (TMS/GMS) and machine translation engine training.”
The guide may in fact be a useful resource for anyone involved in TM management: from the seasoned translation buyer or LSP project manager in need of a refresher, to a new localization coordinator looking for a crash course in TM management.
Slator 2019 Neural Machine Translation Report: Deploying NMT in Operations
32 pages, NMT state-of-the-art, 5 case studies, 30 commentaries, NMT in day-to-day operations
The guide is organized into a selection of topics containing practical tips highlighting key business considerations as well as specific recommendations. It touches on a range of topics including the role and importance of TM champions (admins), how to make the most of your TM, e.g. through applying tags, removing inconsistencies and obsolete/duplicate content, how to decide which number of TMs you actually need, and how to keep your TMs in ship-shape with general housekeeping.
A Few Teasers
Insights: Case studies and community feedback from the 32 survey participants also offer some food for thought. Did you know for example that:
- 78% of respondents group their TMs by product family, business unit or code stack, and
- Most use tags (metadata) to indicate product and data type among other things.
Self-serve: The self-service nature of the guide means that readers can apply recommendations that make sense based on their own specific business context, e.g.
- “You may need to separate TMs based on intended use (desktop vs mobile),” and
- “If the same style guide is used to translate several different components, it may make sense to group all such content into one TM.”
MT vs TMS: The guide also takes the intended use of the TM content into account, helpfully making specific recommendations based on whether the content will be used for TMS or MT engine training, e.g.
- For TMS but not MT engine training, the recommendation is to detect and fix technical issues with the content, and
- Normalizing quotation marks is recommended for MT training only.
Click here to peruse the Best Practices in Translation Memory Management guide.
