πŸ‘¨β€πŸŽ“About Me

Yongmin Yoo (유용민) is an AI researcher and PhD candidate at Macquarie University, affiliated with the Frontier AI Research Centre. His research spans mechanistic interpretability and trustworthy LLMs, structured document intelligence, and AI-driven patent intelligence. By combining his academic background in law and industrial engineering with research and industry experience in NLP, he develops reliable and interpretable language technologies for high-stakes legal and technical domains.

Before pursuing his PhD, he worked as an NLP researcher at NHN for three years. He holds a master’s degree in Industrial Engineering from Inha University and bachelor’s degrees in Law and Industrial Engineering from Kyungsung University, providing a distinctive interdisciplinary foundation for his research at the intersection of AI, technology, and law.

He is one of the few interdisciplinary AI researchers with formal degrees in both law and industrial engineering, complemented by doctoral training in computing and professional experience in NLP research and development. His distinctive strength lies in his ability to understand legal principles and institutional requirements while critically examining the inner workings and technical limitations of modern language models. Drawing on this rare combination of expertise, he bridges law and NLP to develop trustworthy and interpretable AI for high-stakes legal and technical domains.

πŸ” Research Interests

01

Mechanistic Interpretability and Trustworthy LLMs

Understanding how language models internally represent and process information to improve their robustness, reliability, and transparency in high-stakes domains.

02

Structured Document Intelligence

Developing reliable methods for generating, representing, and evaluating complex legal and technical documents with hierarchical structures.

03

AI for Patent Intelligence and Decision Support

Applying language models and machine learning to patent analysis, evaluation, similarity assessment, valuation, and innovation strategy.

πŸ’‘ News

Sep 2026
Paper accepted at AACL-IJCNLP 2026 πŸŽ‰πŸŽ‰
Corresponding Author
Domain-Agnostic Neural Topic Modeling with Contextual Token-Level Semantic Graph Representation
View paper
Aug 2026
Paper accepted at EMNLP 2026 πŸŽ‰πŸŽ‰
First Author
Adaptive Cost-Efficient Evaluation for Reliable Patent Claim Validation
View paper
Apr 2026
Paper accepted at ACL 2026 πŸŽ‰πŸŽ‰
First Author
PatentMind: A Multi-Aspect Reasoning Graph for Patent Similarity Evaluation
View paper
Nov 2025
Paper accepted at EMNLP 2025 πŸŽ‰πŸŽ‰
First Author
PatentScore: Multi-Dimensional Evaluation of LLM-Generated Patent Claims
View paper

πŸ’» Service

Conference Organization & Leadership

Nov 2026
Special Session Chair at BESC 2026
β€œIntegrated Smart Systems and Data-Driven AI”

Program Committee (PC) Member

2024

Journal Reviewer