Watson Language Translator | What Language Barrier?
“If you talk to someone in a language they understand, that goes to their head. If you talk to them in their language, that goes to their heart”. — Nelson Mandela
English, often known as the lingua franca of business, is used in international organizations and has the status of official language in over 20 countries. While accepting English as a standard is simple, that simplification leads to a missed opportunity for organizations that have their content limited to one particular language.
It’s easy to assume that offering services in a single language may be best for a company given the high cost of translators, reviewers, and domain specificity; but companies may be limiting themselves from new markets by constraining content to a few languages. Due to increased globalization, there are now “languages of business” and the importance of multi-lingual content in today’s society is paramount.
Watson Language Translator
Given the increasing importance of being multilingual, Watson Language Translator is a service to help you expand your content and reach new markets. The service allows users to translate text from one language to another. Available in API form, anyone may to embed this functionality to translate applications, websites, documents, chats, and more!
The service comes pre-trained for immediate use out of the box; however, you are are to train the service to allow domain specific customizations. What do I mean by this? You are able to train Watson Language Translator for specific domains or industries.
Custom Models
A common pain-point when using any translator is domain specific translations.
We have the English documentation available, but it is hard to find a good translation service for my specific domain.
Say you are working in the telecommunications industry, words such as ‘frequency’ or ‘signal’ might have specific meanings or different word associations. With Watson Language Translator, you are able to include a few customization capabilities.
Forced Glossary
Proper nouns, company names, or direct translations are phrases you may want translated in a certain way. If you want specific control over translation behavior, use a forced glossary. In the example below, you see a proper noun that should not be translated and a better suited word for automotive headlights.
Examples from English to French:
- International Business Machines > International Business Machines
- Headlights > Feux de croisement
Corpus Customization
Looking for more customization than a forced glossary? Pair it with corpus customization. Using a parallel corpus will provide more translations for the base model to learn from.
These phrases or text snippets, added in a .tmx file, from industry specific documents will allow the resulting custom model to translate text based on the original model and the one you trained.
Try the Demo!
What’s Coming?
From sample applications, to launching our upcoming Neural Machine Translation Engine, there are a lot of things to get excited about with the IBM Watson Language Translator.
Learn More & Get Started:
Watson Language Translator | Neural Machine Translation | Getting Started | Sample Apps
Zia Mohammad is on the product management team for Watson Language Translator and Watson Natural Language Classifier. His passions center around: AI, emerging technologies, space, and neuroengineering.
Feel free to comment or reach out for more information!
