About Me

I am Jing Zhu, a first year PhD student at the University of Michigan. My advisor is Danai Koutra. My primary research interest lies in knowledge graphs.



NegatER: Unsupervised Discovery of Negatives in Commonsense Knowledge Bases. [link]

Tara Safavi Jing Zhu, Danai Koutra
The 2021 Conference on Empirical Methods in Natural Language Processing (EMNLP 21, acceptance rate 25.6%)
  • We propose NegatER, the first Unsupervised framework that ranks potential negatives in commonsense KBs using a contextual language model (LM).
  • Node Proximity Is All You Need: A Unified Framework for Proximity-Preserving and Structural Node and Graph Embedding. [link]

    Jing Zhu*, Xingyu Lu*, Mark Heimann, Danai Koutra
    SIAM International Conference on Data Mining (SDM 21, acceptance rate 21%)
  • We present the first unified framework of node embedding, Nonlinear Proximity Matrix Embedding (NPME) that includes both proximity-preserving and structural embeddings.
  • Selected Awards



    Deep Fuzzy Graph

    Advisor: Qiaozhu Mei
    Research project, Foreseer Group
  • Reproduced DeepGraph in Pytorch.
  • Optimized DeepGraph using multiple Gaussian Functions which mean and variance can be learned by Convolutional Neural Network.
  • On the Transferability of NeRF Representations

    Jing Zhu, Daniel Geng, Sarah Jabbour, Jung Min Lee
    Instructor: David Fouhey
    Course Project, EECS 542, 2020 Fall.
  • We investigated if NeRF representations learned from one scene can be helpful to learning presentations for similar scenes
  • Twitter Semantic Analysis for COVID-19

    Yuhang Zhou, Jing Zhu
    Instructor: Danai Koutra
    Course Project, EECS 476, 2020 Winter.
  • Hydrated tweets related to COVID-19 from Twitter API. Applied emotion recognition for each tweet and analyzed sentiments in each twitter community.
  • [report]