Sign In

Communications of the ACM

ACM TechNews

A Translator Tool With a Human Touch

View as: Print Mobile App Share:
IBM researchers David Lubensky and Salim Roukos

IBM researchers David Lubensky, left, and Salim Roukos are using the company's 400,000 workers to improve digital translation.

Credit: Ari Fishkind / IBM

Nearly 100 IBM researchers are working on n.Fluent, a project to create an automatic translation tool that is fast and accurate enough to be used in instant-messaging between speakers using two different languages.

The n.Fluent project aims to teach the computer terminology that is specific to IBM's business and allow the computer to learn where it went wrong in previous translation efforts. The researchers are extracting and organizing contributions from IBM's 400,000-member workforce, which is distributed across more than 170 countries.

IBM recently issued a "worldwide translation challenge" to its employees, in which about 6,000 IBM employees made improvements in 11 languages to more than 2 million words of text translated by n.Fluent. The submitted data was used to update the translation models. The n.Fluent project is an example of IBM's use of social networking and employee crowdsourcing, which also has been applied to IBM's version of Wikipedia, called Bluepedia, which has contributions from 1,300 employees.

Another internal IBM social networking tool is Dogear, which is modeled after Delicious and enables employees to share links and tagged items from the Internet and the company's intranet. Dogear has become more popular than IBM's own internal search engine, says the IBM Center for Social Software's Irene Greif.

From The New York Times
View Full Article


Abstracts Copyright © 2009 Information Inc., Bethesda, Maryland, USA


No entries found

Sign In for Full Access
» Forgot Password? » Create an ACM Web Account