Clark Mingxuan Ju

Clark Mingxuan Ju

Research Scientist at Snap Inc.

About me

I am a research scientist in the User Modeling & Personalization team at Snap Research. My primary focuses include generative recommendation and representation learning in general.

Before joining Snap, I earned a Ph.D. degree in Computer Science and Engineering from the University of Notre Dame in 2024, advised by Dr. Fanny Ye. Prior to my Ph.D., I received my B.S. and M.S. from Case Western Reserve University, supervised by Dr. Soumya Ray with a focus on machine learning.

News

[2025.05] Two papers are accepted to the research track at KDD'25. One paper studies the popularity bias of recommender systems and the other paper studies cross-domain sequential recommendation. See you in Toronto.

[2025.05] One paper that studies backward-compatible embedding learning has been accepted to ICML'25. Congrats to Ngoc!

[2025.04] One paper that studies learning universal user representations leveraging cross-domain user intent has been accepted to the industrial track of SIGIR'25. Thanks for the joint efforts from scientists and engineers at Snapchat. See you in Italy.

[2025.01] Two papers about RecSys are accepted to WWW 2025 and selected as oral presentations. One studies the learning dynamic of RecSys models from the perspective of matrix rank to accelerate the training process; the other studies RecSys model parameter reduction using graph-based hashing methods. Congrats to Donald and Xinyi!

Preprint(s):

  • Enhancing Item Tokenization for Generative Recommendation through Self-Improvement
    R. Chen, M. Ju, N. Bui, D. Antypas, S. Cai, X. Wu, L. Neves, Z. Wang, N. Shah, T. Zhao
    arXiv [pdf]

  • Heuristic Methods are Good Teachers to Distill MLPs for Graph Link Prediction
    Z. Qin, S. Zhang, M. Ju, T. Zhao, N. Shah, Y. Sun
    arXiv [pdf]

  • Beyond Unimodal Boundaries: Generative Recommendation with Multimodal Semantics
    J. Zhu, M. Ju, D Koutra, N. Shah, T. Zhao
    arXiv [pdf]

  • One Model for One Graph: A New Perspective for Pretraining with Cross-domain Graphs
    J. Liu, H. Mao, Z. Chen, W. Fan, M. Ju, T. Zhao, N. Shah, J. Tang
    arXiv [pdf]

  • Harec: Hyperbolic Graph-LLM Alignment for Exploration and Exploitation in Recommender Systems
    Q. Ma, M. Yang, T. Zhao, N. Shah, R. Ying
    arXiv [pdf]

Selected Publications

[Full List] [Google Scholar]

  • Revisiting Self-attention for Cross-domain Sequential Recommendation
    M. Ju, L. Neves, B. Kumar, L. Collins, T. Zhao, Y. Qiu, C. Dou, S. Nizam, S. Yang, N. Shah
    KDD 25 [pdf] [code]

  • Learning Universal User Representations Leveraging Cross-domain User Intent at Snapchat
    M. Ju, L. Neves, B. Kumar, L. Collins, T. Zhao, Y. Qiu, C. Dou, Y. Zhou, S. Nizam, R. Ozturk, Y. Liu, S. Yang, M. Malik, N. Shah
    SIGIR 25 [pdf]

  • Understanding and Scaling Collaborative Filtering Optimization from the Perspective of Matrix Rank
    D. Loveland, X. Wu, T. Zhao, D. Koutra, N. Shah, M. Ju
    WWW 25 [pdf]

  • How Does Message Passing Improve Collaborative Filtering?
    M. Ju, W. Shiao, Z. Guo, Y. Ye, Y. Liu, N. Shah, T. Zhao
    NeurIPS 24 [pdf] [code]

  • GraphPatcher: Mitigating Degree Bias for Graph Neural Networks via Test-time Augmentation
    M. Ju, T. Zhao, W. Yu, N. Shah, Y. Ye
    NeurIPS 23 [pdf] [code]

  • Multi-task Self-supervised Graph Neural Networks Enable Stronger Task Generalization
    M. Ju, T. Zhao, Q. Wen, W. Yu, N. Shah, Y. Ye, C. Zhang
    ICLR 23 [pdf] [code]

  • Let Graph be the Go Board: Gradient-free Node Injection Attack for Graph Neural Networks via Reinforcement Learning
    M. Ju, Y. Fan, C. Zhang, Y. Ye
    AAAI 23 [pdf] [code]

  • Grape: Knowledge Graph Enhance Passage Reader for Open-domain Question Answering
    M. Ju*, W. Yu*, T. Zhao, C. Zhang, Y. Ye
    EMNLP 22 (Findings) [pdf] [code]

  • Adaptive Kernel Graph Neural Network
    M. Ju, S. Hou, Y. Fan, J. Zhao, Y. Ye, L. Zhao
    AAAI 22 [pdf] [code]

* stands for equal contribution.

Contact

  • Snap Email: mju [at] snap [dot] com
  • ND Alumni Email: mju2 [at] alumni [dot] nd [dot] edu
  • Location: 110 110th Ave NE, Bellevue, WA