About
I am a Ph.D. student at the KAIST School of Computing, working under the supervision of Prof. Alice Oh. My research focuses on learning representations of structured and unstructured knowledge using Graph Neural Networks (GNNs) and Large Language Models (LLMs).
Specifically, I studied graph representation learning methods to leverage pairwise and higher-order interactions for graph-structured data (edges [C2], partial subgraphs [C3], subgraphs [C4], and k-hop subgraphs [P1]).
Currently, my research explores the intersection of graph and language models, with a focus on uncovering latent structures in unstructured language data. My ongoing works are integrating GNNs and LLMs to develop multi-cultural network and language models [W4], searching gene interactions regulating Multiple Sclerosis from single-cell RNA-seq data, and analyzing U.S. lobbying networks.
Recent Publications (See all)
Dongkwan Kim, Junho Myung, and Alice Oh. "Salad-Bowl-LLM: Multi-Culture LLMs by In-Context Demonstrations from Diverse Cultures", Workshop on Socially Responsible Language Modelling Research at NeurIPS (NeurIPS SoLaR), 2024
Chani Jung, Dongkwan Kim, Jiho Jin, Jiseon Kim, Yeon Seonwoo, Yejin Choi, Alice Oh, and Hyunwoo Kim. "Perceptions to Beliefs: Exploring Precursory Inferences for Theory of Mind in Large Language Models", Empirical Methods in Natural Language Processing (EMNLP), 2024
Dongkwan Kim and Alice Oh. "Generalizing Weisfeiler-Lehman Kernels to Subgraphs", Arxiv, 2024
Dongkwan Kim and Alice Oh. "Translating Subgraphs to Nodes Makes Simple GNNs Strong and Efficient for Subgraph Representation Learning", International Conference on Machine Learning (ICML), 2024
Dongkwan Kim, Jiho Jin, Jaimeen Ahn and Alice Oh. "Models and Benchmarks for Representation Learning of Partially Observed Subgraphs", International Conference on Information and Knowledge Management (CIKM, Short Papers Track), 2022
Dongkwan Kim and Alice Oh. "Efficient Representation Learning of Subgraphs by Subgraph-To-Node Translation", Workshop on Geometrical and Topological Representation Learning at ICLR (ICLR GTRL), 2022
Dongkwan Kim and Alice Oh. "How to Find Your Friendly Neighborhood: Graph Attention Design with Self-Supervision", International Conference on Learning Representations (ICLR), 2021
Education
- M.S. School of Computing, KAIST, Sep 2019
- B.S. Major in Computer Science and Minor in Chemistry, KAIST, Feb 2018
Talks & Presentations
Dongkwan Kim. "Salad-Bowl-LLM: Multi-Culture LLMs by Mixed In-Context Demonstrations", International NLP Workshop at KAIST 2024, 2024
Dongkwan Kim. "Leveraging Structure for Graph Neural Networks", IBS Data Science Group Seminar, 2022
Dongkwan Kim. "How to Find Your Friendly Neighborhood: Graph Attention Design with Self-Supervision", KAIST AI Workshop 21/22, 2022
Dongkwan Kim. "How to Find Your Friendly Neighborhood: Graph Attention Design with Self-Supervision", Learning on Graphs and Geometry Reading Group (LoGaG), 2021
Academic Services
- Reviewer: ICLR (2020, 2022, 2024, 2025), GRL+ Workshop (2020), ACL ARR (2021/09, 2024/06), GTRL Workshop (2022), NeurIPS (2022, 2023, 2024), LoG (2022, 2023), ICML (2023), GLB Workshop (2023)
- Student volunteer: ICLR Social ML in Korea (2020), ICLR (2021), NeurIPS 2022 at KAIST (2022), NYU-KAIST Talk Series on LMs (2023)
- Organizer: KAIST AI Workshop (21/22), International NLP Workshop at KAIST (2024)
- Contributor: KAIST ILP Tech (2022/03)
Teaching Experiences
- TA, Head TA of Data Structure (Spring 2018, Fall 2018)
- Head TA, TA of Machine Learning for Natural Language Processing (Fall 2019, Spring 2021), Best TA Award at Fall 2019
- Head TA of Deep Learning for Real-world Problems (Spring 2020, Fall 2020), Best TA Award at Spring 2020
- TA of AI Tech Boostcamp at NAVER Connect Foundation (Fall 2024)