MicroTalks
Track: Inside Track | IN8 | SPLIT SESSION |
Wednesday, July 13, 2022, 5:00am – 5:45am
Held in: Tiergarten
Presenters:
Wouter Maagdenberg - TXTOmedia
Marina Pantcheva - RWS
Min Tan - Google
Anna Zaretskaya - TransPerfect
Host: Daniel Goldschmidt
This session will offer several quick and dynamic presentations covering topics and ideas too interesting to ignore:
Why Size Matters: How Your LQA Sample Size Influences the Perception of Translation Quality – Marina Pantcheva (RWS)
The industry-standard metric for measuring translation quality disregards the size of the reviewed sample: 1 error in a 100-word sample is just as bad as 10 errors in a 1000-word sample. This tenet is challenged by linguistic theory, according to which the sensitivity to language rule violations is nonlinear. We set out to validate this theory against data and explore the dependency between error counts, sample sizes, and the overall quality experience ratings in real-live localization quality assurance (LQA) reviews. The result was surprising, but in line with the theory: reviewers become less sensitive to translation errors as the reviewed sample size decreases.
Takeaways: Attendees will learn how the industry standard LQA measures are designed; and they will be introduced to a novel approach to LQA scoring that has the potential to solve the long-standing problem of how to accurately calculate LQA scores on small review samples, while at the same time reflecting the actual linguistic experience of users themselves.
Inclusive Localization as a Company Strategy – Min Tan (Google)
There are different elements in inclusive product design, including gender equality, racial equality, cultural nuances, religion, and so on. Within a product’s design to launch, there are multiple touch points including user acquisition, product development, and localization. In this presentation, we will lay out the fundamental elements of inclusive localization and best practices for gaming, software and apps, and ecommerce companies.
Takeaways: Attendees will learn about inclusive product design core elements; best practices; and dos and don’ts.
Localized Videos Unlimited – Wouter Maagdenberg (TXTOmedia)
Localizing videos is often still an afterthought and a balancing act. It shouldn’t be. Especially when dealing with educational content such as training, instruction, and how-to videos since native language is key for understanding and remembering. Video localization after production is unnecessarily complex, consuming extra time and money. It’s better to take care of it at once while creating the original. How? Well, think in video fragments, not in titles, and see video as an integral part of your publication flow, like print and pdf. It stimulates consistency, compliance and even speeds time-to-market.
Takeaways: Attendees will learn that localized videos are key for learning; fragment-based video from the start is the best practice; and leverage existing content as the script.
Using AI to Generate Product Content – Anna Zaretskaya (Transperfect)
AI is gaining more space in the content life cycle. Machine translation (MT) has become a standard tool in localization, for almost all verticals and languages, and now we have started looking beyond MT. Today, AI can solve many text-related tasks extremely well, and among them is text generation. In this session, we will discuss a specific use case of applying neural language models to generation of product descriptions in the retail domain. Language models, the most well-known being GPT3, when prompted correctly, can generate text nearly indistinguishable from text written by humans. This has clear implications: descriptions of products can be created much faster and cost-efficiently than simply writing them from scratch. We will share our experience in this field and the common problems our clients face, such as controlling the quality of the output, accounting for SEO, and finally making the content multilingual.
Takeaways: Attendees will get familiar with AI for text generation; be able to decide if it can be beneficial for their organization; and get an understanding of the common challenges in such implementation and how to face them.