Human Language Technology Research Institute Established at UTD

RICHARDSON, Texas (Feb. 25, 2002) – The University of Texas at Dallas (UTD) today announced that it has established a research institute aimed at advancing the understanding and uses of natural language processing, a science that has grabbed the attention of federal government intelligence agencies in the wake of Sept. 11.

The technology, which utilizes sophisticated software tools, has the potential to help security personnel simultaneously screen and analyze millions of voice and text communications in an attempt to uncover terrorist or other criminal activities. Other possible uses include customer support applications in many lines of business.

The research, new at UTD, will be carried out at the newly established Human Language Technology Research Institute in the Erik Jonsson School of Engineering and Computer Science. The effort will be headed by Dr. Sanda Harabagiu, director of the institute and an associate professor of computer science in the Jonsson School. Harabagiu also holds the Jonsson School Research Initiation Chair Professorship.

Harabagiu joins UTD from The University of Texas at Austin, where she was a member of the computer science faculty. An expert in natural language processing, knowledge processing and artificial intelligence, Harabagiu has conducted research on information extraction from large, complex databases at U.T. Austin, SRI International, the University of Southern California and Southern Methodist University.

“The establishment of the Human Language Technology Research Institute provides UTD entry into an exciting field that has possible widespread implications for our nation’s security,” said Dr. Andrew J. Blanchard, executive director of industry-university relations and senior associate dean of the Jonsson School. “Dr. Harabagiu’s arrival will broaden the scope and quality of our research and educational opportunities and further position the Jonsson School as one of the leading schools of engineering and computer science in the United States.”

The institute has been established with approximately $3 million in initial funding, which comes from the university and a variety of other organizations, including the National Science Foundation. As the faculty and staff of the institute grow, so, too, will opportunities for additional, sizeable grants, according to Harabagiu.

“Since Sept. 11, there has been a heightened interest in this work among federal agencies and, consequently, millions of dollars are being made available for research of the kind we will be doing at UTD,” she said.

Harabagiu and her colleagues will develop sophisticated software tools that can quickly and reliably process, identify, analyze and extract desired information from huge collections of documents, composed of both text and speech.

“Most people are aware of how a search engine works on the Internet,” Harabagiu explained. “You type in a key word or words and it returns multiple – sometimes hundreds or thousands – word matches found on Web sites or other online documents.

“Human language technology is far more sophisticated and precise. For example, I can pose a specific question, asking what poet wrote a particular poem, and, by querying the Internet or other database, get an immediate, specific – and correct – answer.

“The difference between the search engine and human language technology is a result of a degree of artificial knowledge or intelligence incorporated into the latter. We are not simply mining data, but knowledge, from assembled information.”

While Harabagiu is an expert on textual question and answering, the new institute also will research the same process involving the spoken word. She is currently interviewing potential researchers with that particular expertise.

When fully staffed, the institute will be composed of more than 10 faculty members and at least 40 doctoral students, Harabagiu estimated.

Her goal for the institute is to become a “center of excellence in human language technology, on par with institutions involved in similar research at Carnegie Mellon University, the University of Pennsylvania and Stanford University.”

The establishment of the institute drew praise from the head of a Dallas-based business that provides speech-enabled communications, collaboration and e-business applications to network service providers and companies around the world.

“We are pleased to serve on the advisory board for the Human Language Technology Research Institute,” said David W. Brandenburg, CEO and chairman of InterVoice-Brite and member of the UTD development board. “UTD’s thrust into human language technology offers InterVoice-Brite exciting opportunities to collaborate with leading experts to develop the next generation of interactive voice solutions that are more conversational, easier to deploy and commercially viable.”