H2 at Unbabel: We’ve been working diligently behind the scenes on product innovation to get new features and solutions into the hands of our customers worldwide.
We’ve sharpened our focus this past quarter on enhancing our workflow integration efforts, by investing more in quality, customization, and localization.
As an extension of Unbabel’s existing features, we built the per-client anonymization feature to help clients maintain the privacy of their users’ personal information. Our customers now have the ability to specify what type of information — whether it be home address, credit card, etc. — should be hidden from our machine and human translators, and remain confidential.
Since not every localization pipeline requires the assistance of human editors, we have further enhanced our machine translation engines to automatically identify and localize content — with the release of Programmatic Localization. This new function allows us to automate the translation process to increase efficiency and reduce turnaround time. What does this look like? It could be as simple as automatically reformatting a date (DD/MM, MM/DD) or the reading direction (from left to right) when translating between languages like English and Hebrew.
We can now also localize content based on an organization’s unique preferences with Custom Localization. This feature’s ability to make translations specific to a market or culture, and according to our customer’s brand, introduces a new level of customization to localization. By giving customers the power to specify how a given Named Entity should be translated and/or formatted, we’re able to push the localization boundaries for certain categories of content. A good example of this is demonstrated when a customer requests a Russian to English translation but asks that Russian Names not be translated so the format remains in the Cyrillic alphabet.
Richer Named Entities
In order to localize Named Entities, they need to be accurately identified. With that in mind, we’ve expanded our Named Entity Recognition engines to detect measurements and numerical units, as well as enhanced existing categories such as names, products, services, and organizations to continuously provide high-quality translation and localization performance.
Enhanced Translation Performance
The contribution of Unbabel editors is a key value proposition to the Unbabel offering, which is why we put emphasis on making their interactions with our editor interface more efficient. With the release of Concordance Search and Translation Alternatives, we can help editors work faster and more accurately. How is this done?
Concordance Search: Unbabel editors can now refer to past translations of specific terms to ensure consistency and accuracy while they work. This feature decreases translation errors and improves the overall quality of Unbabel translations.
Translation Alternatives: This feature accounts for instances where there is no direct translation available from one language to another. The tool provides Unbabel editors access to alternative translations that display partially-matching suggestions when exact translations are limited. Our goal is to increase the rate at which translation alternatives are accepted by editors to 70% to improve translation quality and speed up the translation process.
Portal Help Widget
With the addition of the Help Widget as a feature of the Unbabel Portal, we are able to provide a simple place for customers to seek help and questions. The Help Widget is linked to where our customers work to provide convenient access to:
Unbabel Customer Happiness team
Unbabel Help Center
Updates on the status of all Unbabel systems
If you have any questions about what we’ve been working on, and how it can help your business improve its multilingual CX, contact us here: