Communications of the ACM
Stanford System Combines Software With Human Intelligence to Improve Translation
Researchers at Stanford University say a system they have designed will help translate text faster and more accurately than is now possible.
Credit: Stuart Miles/Shutterstock
Stanford University researchers have developed a system designed to help people translate text faster and more accurately than is currently possible.
The system relies on a hybrid approach that merges human and machine intelligence, and it hopes to enter the $34-billion-per-year global market for professional translation services.
Stanford computer science graduate student Spence Green and Christopher Manning, a professor of linguistics and computer science, presented their hybrid approach to translation at two computer science conferences in October. "Our system augments the human translator and increases efficiency, accuracy, and productivity," Green says.
The system involves providing human translators with an interface that resembles a word processor. When the translator begins typing a translation, the system offers suggestions for key words and phrases that could have multiple meanings. The system is able to revise its suggestions in real time as the translator works, fine-tuning the software and improving the quality of its assistance.
At the conferences, the researchers discussed how the system's interface design provides higher-quality final translations and how feedback from human translators is used to correct mistakes and even adapt translations to their style.
From Stanford Report (CA)
View Full Article
Abstracts Copyright © 2014 Information Inc., Bethesda, Maryland, USA
No entries found