NSF CAREER Award Boosts Data Extraction Research, Voice-Tech Research
University of Texas at Dallas faculty members in the Erik Jonsson School of Engineering and Computer Science received 2024 Faculty Early Career Development Program (CAREER) awards from the National Science Foundation (NSF) to support research to reduce information overload and advance voice technology.
Since 2010, UT Dallas faculty members have received 58 NSF CAREER awards, which support early-career faculty who have the potential to serve as academic role models in research and education and to lead advances in their fields.
Dr. Xinya Du, assistant professor of computer science and Dr. Berrak Sisman, assistant professor of electrical and computer engineering, both received the five-year grants this year.
Du’s $561,000 CAREER grant supports his work to build methods to extract key information from large amounts of text.
The vast amounts of event-related information published daily create an overwhelming flood of data that far exceeds the cognitive capacity of any individual, Du said.
“My research focuses on developing advanced techniques to automatically extract succinct event knowledge, enabling people to quickly capture critical insights from complex documents,” Du said. “I aim to tackle the information overload problem in people’s daily lives by developing innovative natural-language processing techniques that transform unstructured text in long and complex documents into structured, comprehensive knowledge graphs.”
Sisman received a $564,000 award to study how factors including background noise, emotions and cultural differences affect a voice recognition system’s ability to recognize and accurately render an individual’s speech.
The project also aims to detect speech spoofing, which involves the fraudulent use of speech samples to impersonate a person’s voice, Sisman said.
“The information will be used to create voice technology that can be used for safeguarding ethical and responsible use of voice generation, for detecting and preventing fraud, and to make it more challenging for unauthorized users to mimic speakers,” she said.