Rui Zhang
Pennsylvania State University, PA·Computer Science·Early-career
People increasingly ask large language models to summarize documents so they can grasp key information and make decisions. But these systems have been criticized on three fronts: they ignore what a particular user wants in a summary, they can overlook the views of some social groups, and they make factual errors.
- Fields
Computer Science
Engineering
- Topics
text summarization
large language models
generative AI
- Grants
- 1 known grant, through Aug 2029
What they work on
- Develop controllable summarization models that satisfy fine-grained, compositional user preferences for both the content and the style of a summary, along with rigorous task definitions and evaluation metrics.
- Build methods that generate fair and unbiased summaries covering diverse social perspectives and conflicting opinions, so that views of particular groups are not neglected.
- Create summarization models that honor human knowledge by handling multimodal sources together: unstructured text, structured tables and citations, and visuals such as figures and plots.
- Deploy an integrated, interactive and visualizable summarization platform, with public benchmarks, to support efficient and accessible document understanding by real users.
More about the research
This project aims to make summarization trustworthy by putting people at the center of how such systems are designed, built and deployed, along three dimensions: controllability for user preferences, fairness across social perspectives, and factuality with respect to human knowledge.
Why it matters
Summaries from language models now feed real decisions in areas such as scientific discovery, legal and democratic processes, and public health. Tools that follow a user's stated needs, represent conflicting viewpoints fairly, and stay faithful to source documents and tables could reduce the risk of biased or factually wrong summaries driving those decisions, and the public benchmarks and interactive platform would give other researchers and practitioners a shared way to measure progress.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Closest labs
Fei Liu
Emory University, GA·Computer Science·Early-career
Uncover the hidden structure of informal, verbose transcripts of spontaneous speech, using deep neural networks and linguistic structure prediction to infer how topics and segments of a conversation relate to one another.
Lu Lin
Pennsylvania State University, PA·Computer Science
Identify critical LLM safety concerns, gaps in current evaluation and infrastructure needs through structured surveys and interviews with experts across computing disciplines.
Anthony Sicilia
West Virginia University, WV·Computer Science
Develop a Probably Approximately Correct (PAC) learning theory that measures how user disagreement, suggestions and topic shifts change the calibration of an AI system's confidence during a conversation.
The school
Pennsylvania State University: 60 labs on PI Pages, $1.3B a year on research, 743 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.