Jingtao Zhan

My name is Jingtao Zhan (詹靖涛, in Chinese). I am now a forth year PhD student at Department of Computer Science and Technology, Tsinghua University. My supervisors are Prof. Shaoping Ma and Prof. Yiqun Liu.

My research interests lie in building intelligent information systems, which primarily comprise two parts.

  • One part of my work focuses on designing optimization algorithms to construct efficient information retrieval systems. This includes designing evaluation schemes, building training frameworks for models, and optimizing index structures. One of my works received the Best Paper Award at WSDM 2022.
  • The other part of my work involves researching how to leverage user interaction logs in AIGC systems to enhance system performance. The motivation for this research stems from the fact that AIGC systems, like a black box, derive their performance mainly from the user’s ability to “cast spells” (prompting). Therefore, I am very interested in how to utilize user interaction logs to unearth the capabilities of AIGC systems and, in turn, further improve system performance. One of my works in this area has been accepted by SIGIR 2024.

I also served as a reviewer and PC member for SIGIR, ACL, theWebConf, WSDM, CIKM, and COLING.

Publications

  • Capability-aware Prompt Reformulation Learning for Text-to-Image Generation.
    Jingtao Zhan, Qingyao Ai, Yiqun Liu, Jia Chen and Shaoping Ma.
    The 47th International ACM SIGIR Conference on Research and Development in Information Retrieval. (SIGIR 2024). [paper] [code]

  • Scaling Laws For Dense Retrieval.
    Jingtao Zhan*, Yan Fang*, Qingyao Ai, Jiaxin Mao, Weihang Su, Jia Chen and Yiqun Liu. (* Equal contribution)
    The 47th International ACM SIGIR Conference on Research and Development in Information Retrieval. (SIGIR 2024).

  • Combining Multiple Supervision for Robust Zero-shot Dense Retrieval.
    Yan Fang, Qingyao Ai, Jingtao Zhan, Yiqun Liu, Xiaolong Wu, Zhao Cao.
    The 38th Annual AAAI Conference on Artificial Intelligence. (AAAI 2024). [paper]

  • Constructing Tree-based Index for Efficient and Effective Dense Retrieval.
    Haitao Li, Qingyao Ai, Jingtao Zhan, Jiaxin Mao, Yiqun Liu, Zheng Liu, and Zhao Cao.
    The 46th International ACM SIGIR Conference on Research and Development in Information Retrieval. (SIGIR 2023). [paper]

  • Disentangled Modeling of Domain and Relevance for Adaptable Dense Retrieval.
    Jingtao Zhan, Qingyao Ai, Yiqun Liu, Jiaxin Mao, Xiaohui Xie, Min Zhang, and Shaoping Ma.
    (ArXiv). [paper] [code]

  • Evaluating Interpolation and Extrapolation Performance of Neural Retrieval Models.
    Jingtao Zhan, Xiaohui Xie, Jiaxin Mao, Yiqun Liu, Jiafeng Guo, Min Zhang and Shaoping Ma.
    The 31th ACM International Conference on Information and Knowledge Management. (CIKM 2022). [paper] [code]

  • Learning Discrete Representations via Constrained Clustering for Effective and Efficient Dense Retrieval.
    Jingtao Zhan, Jiaxin Mao, Yiqun Liu, Jiafeng Guo, Min Zhang and Shaoping Ma.
    The 15th ACM International Conference on Web Search and Data Mining. (WSDM 2022 Best Paper Award). [paper] [code]

  • Jointly Optimizing Query Encoder and Product Quantization to Improve Retrieval Performance.
    Jingtao Zhan, Jiaxin Mao, Yiqun Liu, Jiafeng Guo, Min Zhang and Shaoping Ma.
    The 30th ACM International Conference on Information and Knowledge Management. (CIKM 2021). [paper] [code]

  • Optimizing Dense Retrieval Model Training with Hard Negatives.
    Jingtao Zhan, Jiaxin Mao, Yiqun Liu, Jiafeng Guo, Min Zhang and Shaoping Ma.
    The 44th International ACM SIGIR Conference on Research and Development in Information Retrieval. (SIGIR 2021). [paper] [code]

  • RepBERT: Contextualized Text Embeddings for First-Stage Retrieval.
    Jingtao Zhan, Jiaxin Mao, Yiqun Liu, Min Zhang, and Shaoping Ma.
    (ArXiv). [paper]. [code]

  • An Analysis of BERT in Document Ranking.
    Jingtao Zhan, Jiaxin Mao, Yiqun Liu, Min Zhang, and Shaoping Ma.
    The 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval. Short Paper. (SIGIR 2020). [paper] [code]

  • Leveraging Passage-level Cumulative Gain for Document Ranking.
    Zhijing Wu, Jiaxin Mao, Yiqun Liu, Jingtao Zhan, Yukun Zheng, Min Zhang and Shaoping Ma.
    The Web Conference 2020. (WWW 2020). [paper]

Education

  • 2024.3-present, visiting research scholar, UIUC.
  • 2020-present, Ph.D. student, Computer Science, Tsinghua University.
  • 2016-2020, B.E., Electronic Engineering, Tsinghua University.

News

  • 10.2022, I gave an invited talk to Amazon Alexa AI about “Disentangled Modeling of Domain and Relevance for Adaptable Neural Retrieval”. [paper] [code]
  • 09.2022, I gave an invited talk to Alibaba (the DAMO NLP team and OpenSearch team) about “Disentangled Modeling of Domain and Relevance for Adaptable Neural Retrieval”. [paper] [code]
  • 03.2022, I gave invited talks to BIT, Jina.ai, and Baidu about “Effective and Efficient Dense Retrieval via Joint Optimization with Compact Index”. [paper] [code]
  • 02.2022, Our paper “Learning Discrete Representations via Constrained Clustering for Effective and Efficient Dense Retrieval” won WSDM’22 Best Paper Award . [paper] [code]
  • 01.2022, I gave an invited talk to Glasgow IR Group about “Effective and Efficient Dense Retrieval via Joint Optimization with Compact Index”. [paper] [code]

Honors and Awards

  • 2022, Longfor Scholarship
  • 2022, Overall Excellence Scholarship (First Prize), Tsinghua University. (Top 5%)
  • 2022, WSDM’22 Best Paper Award
  • 2021, Overall Excellence Scholarship (Second Prize), Tsinghua University. (Top 10%)
  • 2020, Outstanding Graduate of Beijing
  • 2020, Outstanding Graduate of Tsinghua University
  • 2019, National Encouragement Scholarship
  • 2017, National Encouragement Scholarship