News & Events

2012 CAREER Award Winner Heng Ji: Teaching Computers to Think

December 12, 2012 | Awards, Faculty, Research

In 2010, Dr. Heng Ji was awarded a $527,134 NSF Faculty Early Career Development (CAREER) Award to support her innovative interdisciplinary research on Natural Language Processing (NLP)and Information Extraction (IE). Dr. Ji is an Assistant Professor of Computer Science at Queens College and a member of the doctoral faculty of both Computer Science and Linguistics at the CUNY Graduate Center.

As an undergraduate at Tsinghua University in Beijing, Dr. Ji applied to the math, chemistry, and physics departments, but was instead assigned to the linguistics program.  “I was not very happy about this,” she says.  “It was just not what I had in mind.”  However, during her first few months of classes, a linguistics professor recognized her broad ranging potential and recommended her to a computer scientist, who went on to become her first mentor.

At 17-years-old, Dr. Ji was given a set of lab keys so she could carry out research in both the linguistics and computer science departments at her university.  Dr. Ji earned her Bachelor’s and Master’s degrees in Computational Linguistics at Tsinghua before moving to New York City to pursue a doctoral degree in Computer Science at New York University.

Dr. Ji began teaching at Queens College in 2008, a year after receiving her PhD from NYU.  However, she didn’t always envision herself as a professor; as an undergraduate she assumed she would find a job in a company and work her way up the corporate ladder.  But, halfway through her doctoral study, she decided to switch gears and focus on entering academia, which would allow her to, in a sense, work for herself.

At Queens College, Dr. Ji currently serves as Director of the BLENDER Lab, described as a “cross-lingual, cross-documents, cross-media information extraction and fusion lab,” where she focuses on the development of new technologies that allow computers to understand information from text, video and audio in all languages.  NLP researchers are trying to teach computers to process language like humans, thus vastly increasing our ability to mine information from the web.

Dr. Ji’s lab works on several projects aimed at advancing NLP.  One area in particular is in information extraction.  Often times the computer must first accurately translate the information from one language to another, even when working with a variety of language styles.  Current translation programs fall short of the ideal.  Dr. Ji explains that when the current programs are applied to text from more casual sources like Twitter or Facebook, the programs fail to properly understand the meaning.  Her lab is now focusing on teaching computers a variety of language styles, from the straightforward news article or technical science paper to the informal YouTube comment.

She is also conducting research to enhance computers’ reasoning skills.  Humans intuitively understand that when seeking information on a person, information about that person’s spouse is useful. However, computers  must be programmed to enable them to draw the logical conclusions we take for granted. Internet search requests would be enhanced by this type of computer reasoning, computers could delve deeper into online information than is possible using simple key word searches.

“I love this research area.” Dr. Ji says.  “I think it gives me a really nice balance between the arts and sciences.  It not only needs you to know the computing skills. It also requires that you understand the language.  If you just know how to do the programming, but do not know how to capture the language, you can’t do the work.”

In addition to the CAREER award, Dr. Ji’s current funding includes a U.S. Military Defense Advanced Research Project Agency Award, a U.S. Army Award and a NSF EAGER award.  The work is providing valuable experiences for her lab team and she has high praise for them.  “I’m very lucky,” she says.  “They work very hard.”

Dr. Ji has been awarded all but one grant for which she has applied. “It’s like a snowball.  Once you get some funding then you get students working, then you’re more productive and you publish more papers,” she says.  “The snowball is becoming larger and larger.”