Trustworthy User Modelling and Personalization (TUMP)

The TUMP Lab advances trustworthy user modeling and personalization through cutting-edge AI techniques, human-centric LLM-based user simulation, and rigorous benchmarking via open-source frameworks.

Research Focus:

  • Designing trustworthy recommendation algorithms—robust, diverse, fair, explainable, and privacy-preserving—by applying cutting-edge techniques e.g., deep learning and large language models across various domains including e-commerce, multimedia, location-based social networks, and healthcare.
  • Developing LLM-based user simulators for human-centric evaluation and optimization of recommender systems.
  • Benchmarking recommender systems through rigorous evaluation and fair comparison by advancing open-source libraries such as DaisyRec and DaisyRec-v2.0.

Latest News

[Nov 2026] One paper was accepted by KDD AI4Science Track as Oral Presentation (6.1% of all submissions). Congratulations to all co-authors!
[Nov 2026] One paper was accepted by WSDM 2027. Congratulations to all co-authors!
[Oct 2026] Welcome Minghui Wang to our team as a Visiting PhD student!
[Oct 2026] Professor Sun will serve as the Proceeding Chairs for ACM RecSys 2027!
[Oct 2026] Professor Sun will serve as the Area Chair for The Web Conference 2027!
Join Us