Yoo Yeon Sung

Yoo Yeon Sung

Ph.D. Student in Information Studies

University of Maryland

Biography

I am a fourth-year Ph.D. candidate at the Information Studies College of the University of Maryland, College Park. I work with Jordan Boyd-Graber and Naeemul Hassan. My research interests are broadly in Human-interactable NLP and Responsible AI.

My recent work focuses on multimodal (e.g., video and text) models in Misinformation domain with human-centered perspective (e.g., considering subjectivity). Orthogonally, I focus on dynamic adversarial dataset generation, calibration, and evaluation in the Question Answering and Fact Checking Domain. I strive to understand how current complex models (e.g., LLMs) impact the human-computer arms race in this era, and thereby tackle on-going information crisis from a human-centered perspective.

Interests

  • Multimodal and Dynamic (HITL) Dataset collection
  • Adversarial Calibration of datasets and models
  • Interpretable AI
  • Multimodal misinformation

Education

  • PhD in Information Studies, 2020-Present

    University of Maryland

  • Masters in Industrial Management Engineering, 2017-2019

    Korea University

  • BS in English Language and Literature & Computer Science and Engineering, 2013-2017

    Chung-Ang University

Publications

(2024). ADVSCORE: A Metric for the Evaluation and Creation of Adversarial Benchmarks. Arxiv.

Arxiv

(2023). Not all Fake News is Written: A Dataset and Analysis of Misleading Video Headlines. Empirical Methods in Natural Language Processing 2023. Long.

PDF Arxiv Dataset Code Video

(2020). Topical Keyphrase Extraction with Hierarchical Semantic Networks. Decision Support Systems 2020.

PDF Arxiv