Two USC Viterbi Information Sciences Institute (ISI) researchers have been named Association of Computational Linguistics (ACL) Fellows — the second time two new ACL fellows have come from the same academic institution. The distinction comes on the heels of ISI’s Jerry Hobbs receiving the ACL Lifetime Achievement Award in 2013.
ISI Director of Natural Language Processing and Professor of Computer Science, Kevin Knight
Research Associate Professor of Computer Science, Daniel Marcu
Kevin Knight and Daniel Marcu, both faculty in the Department of Computer Science, have both been recognized for their extensive research into computational linguistics, a blanket term that encompasses everything computers can do with text and speech. The pair is improving machines’ ability to learn translation patterns without needing someone to continuously input rules manually — work that has helped yield results used all over the world, trillions of times every day in everything from Google Translate to Skype’s new dynamic speech translation service.
“Natural language translation is a fairly new area within the computational linguistics field, but the ultimate goal is to use machine learning to essentially knock down every language barrier in the world,” Marcu said. “We believe that most problems in the world exist because of a lack of communication, and as that communication is established across cultures and barriers, the world is becoming a better place.”
Computational linguistics is the scientific study of language from an artificial intelligence perspective with linguists providing either "knowledge-based" or "data-driven" models of various kinds of linguistic phenomena. Both Knight and Marcu work on the data-driven statistical side of the field.
The ACL was founded in 1962 and established the fellows program in 2011. The organization has named 26 fellows to date, including the six named in 2014.
“Both our ACL fellows this year and our lifetime achievement award winner last year demonstrate the uniquely significant impact ISI’s research is having in the area of natural language processing,” said Prem Natarajan, USC ISI executive director. “We are immensely proud of all their achievements and look forward to their continued success and contributions to the field.”
Knight, who also teaches a Ph.D.-level natural language processing class, has worked with documents encrypted either by the military or by virtue of being written in “dead” languages, developing algorithms to read them — a process similar to code breaking. Marcu is credited with re-energizing industry discussions about this application of artificial intelligence and has spent the last 15 years working to perfect machine translation of text and speech, contributing to the dynamic translators available to the general public today.
While they have distinctly different approaches, both researchers appreciate the big picture implications of their work.
“Our dream in translations is that everywhere you go and everything you see or hear should come to you in your own native language,” Knight said. “It won’t matter if it’s Japanese, Chinese or Hindi — invisible computer software should be eliminating that language barrier for you.”
Marcu is equivalently bullish on the widespread possibilities.
“My children are in primary school right now, and I believe that by the time they grow up, they’ll be able to visit other countries and not need to learn those languages,” he said. “Portable devices would enable them to go to China, Japan or anywhere in Europe and have a fantastic tourist experience without having to learn the native language.”