Jonghyun (Jong) Song

Jonghyun (Jong) Song

he/him

Research Interests: Information Retrieval and Multi-modal Lanugage Models

Jonghyun (Jong) Song

Hello! I’m Jonghyun (Jong) Song, a Ph.D. student in Data Science at Seoul National University. Before joining GSDS, I was trained to be a junior engineer at the Department of Mechanical and Aerospace Engineering of the same school (Dept. of MAE, SNU).

My primary research interests include Natural Language Processing, Information Retrieval (IR), and Multi-modal Language Models. I’m particularly fascinated by language model embeddings—the idea of encoding vast information into numerical representations is inspiring. In this regard, I’m exploring how to optimize these embeddings for downstream tasks, specifically focusing on efficient modeling of query-document and document-document interactions in IR systems. Also, I’m interested in advancing the reasoning capabilities of large language models for multi-modal tasks, enabling them to interpret and connect diverse data types more effectively. For more details, please refer to my CV.

In my free time, I like to play tennis :tennis: and watch soccer games :soccer: (I’m a fan of 1. FSV MAINZ 05 in Bundesliga!)

:newspaper: News

  • Jan. ‘25 I have started my internship with the Hyper-Personalized Search Team at NAVER, the leading search engine in Korea!
  • Nov. ‘24 Our paper Comparing Neighbors Together Makes it Easy: Jointly Comparing Multiple Candidates for Efficient and Effective Retrieval is accepted to EMNLP 2024 Main!
  • Aug. ‘24 I presented a spotlight paper talk at the ACL Workshop on Representation Learning for NLP (Repl4NLP) at ACL ‘24!

:bookmark_tabs: Papers and Presentations

  • Jonghyun Song, Cheyon Jin, Wenlong Zhao, Andrew McCallum, Jay-Yoon Lee, Comparing Neighbors Together Makes it Easy: Jointly Comparing Multiple Candidates for Efficient and Effective Retrieval, In EMNLP 2024 (Main)/Spotlight Talk at Repl4NLP workshop in ACL 2024
  • Jonghyun Song, Eunyi Lyou, Redefining Information Extraction from Visually Rich Documents as Token Classification, In IJCAI 2024 Competition of Visually Rich Form Document Intelligence and Understanding (VRDIU)

🏆Honors and Awards

  • Jul. ‘24 2nd Place, VRDIU Competition (Track A) on IJCAI 2024 sponsored by Google Research
  • Nov. ‘23 1st Place (Minister’s Award) on K-Datascience Hackathon by Ministry of Science and ICT, Korea
  • Mar. ‘22 Park Min-Chul Data Science Challenge Scholarship
  • Mar. ‘21 Sanhak (Industrial-Educational Cooperation) Foundation Scholarship

:pencil2: Teaching

  • Fall ‘23 Head TA, Natural Language Processing with Neural Networks, Seoul National University
  • Fall ‘23 Instructor, Big Data Fintech Specialist Training Course, Ministry of Employment and Labor
  • Spring ‘23 Head TA, Math and Statistics Foundations for Data Science, Seoul National University
  • Fall ‘22 Head TA, Applications of Natural Language Processing, Seoul National University
  • Fall ‘22 TA, Data Science Boot Camp, Seoul National University
  • Winter ‘20 Math Tutor, Self-Paced Learning & Tutoring Program, Seoul National University
  • Fall ‘19 Undergraduate TA, Creative Engineering Design, Seoul National University