Text Information Modeling and Analysis Group

The Text Information Modeling and Analysis (TIMAN) group is part of the Data and Information Systems (DAIS) Lab of the Computer Science Department at University of Illinois at Urbana-Champaign. The TIMAN group is primarily composed of the graduate students, undergraduate students, and visitors working with Prof. ChengXiang Zhai. Since the group was founded in 2002, a large number of group members, including 41 Ph.D. students, over 50 M.S. students, and over 40 undergraduate students have graduated. See the complete list of TIMAN alumni for detail.
TIMAN Research Vision

We work on a wide range of research problems on "big data", especially big text data, with a general goal of developing innovative intelligent information systems to augment human intelligence by enabling people to interact with large amounts of data and discover actionable insights/knowledge from the data for optimizing complex decisions and many user tasks (see the illustration).

Our research uses techniques from multiple fields, such as information retrieval, natural language processing, data mining, and machine learning. We are especially interested in developing formal models, general intelligent algorithms, and general tools that can be used to support many different applications as well as using them to build innovative useful application systems in specific domains such as healthcare, medicine, online learning, and scientific discovery.

We emphasize optimization of human-AI collaboration and maximization of the combined intelligence of humans and AI systems. To this end, we study how to mathematically model and simulate users, which is required for both evaluating and training interactive intelligent systems. Leveraging user modeling and simulation, we study how to optimize sequential decisions for an intelligent agent to interact with users to support their tasks in a personalized manner while minimizing a user's overall effort. To enable explainable AI (XAI) and trustworthy AI, we are also interested in studying human-like natural language processing techniques, especially neuro-symbolic models. We apply general models and algorithms to develop specific intelligent task agents in multiple application domains such as healthcare, education, and scientific discovery. See the list of TIMAN projects for details about the projects that we have completed and our current projects.

Our research has been funded by many different sources, including government agencies (e.g., NSF, NIH, DHS, DOI, NASA, AFOSR, IARPA, DOE, FDA), industry (e.g., Microsoft, Google, Yahoo, IBM, HP, LinkedIn, Amazon, Intel), and private foundations (e.g., Alfred P. Sloan and Robert Wood Johnson).