Research Interest 😆

Natural Language Processing, Commonsense Reasoning, Knowledge Editing, Hallucination, Language Resource & Evaluation

Education 🎓

2025 Ph.D in Computer Science at Korea University (Seoul), South Korea
2020 B.A. in Business School at Korea University (Seoul), South Korea
2020 B.A. in English Language and Literature at Korea University (Seoul), South Korea

Honors & Awards 🏆

Top 5 K-AI Team

• Appointed to the National Representative AI Research Team (NC AI consortium)

Best Paper 2024

• The 36rd Annual Conference on Human & Cognitive Language Technology (HCLT2024): Model Interpretability Section

Outstanding Paper Award 2023

• The 35rd Annual Conference on Human & Cognitive Language Technology (HCLT2023): Language Resource and Information Retrieval Section

NAVER PH.D Fellowship Award 2022

• NAVER selects doctoral students who have published excellent papers as the first author in world top-tier conferences or journals with outstanding presentation achievements and awards them scholarships.

1st place in WMT 2022 (Quality Estimation - Critical Error Detection) at EMNLP 2022

• Sugyeong Eo, Chanjun Park, Hyeonseok Moon, Jaehyung Seo

Outstanding Paper Award 2022

• The 34rd Annual Conference on Human & Cognitive Language Technology (HCLT2022): Language Resource and Information Retrieval Section

Ranked 4th on the CommonGen 1.1 Leaderboard Jul. 2021

• PU-GEN: Enhancing generative commonsense reasoning for language models with human-centered knowledge (Knowledge-Based Systems 2022)
  (Nov. 2022 Ranked 7th, CommonGen 1.1)

Outstanding Paper Award 2021

• The 33rd Annual Conference on Human & Cognitive Language Technology (HCLT2021): Language Resource and Question Answering & Speech Section

Invited Talk & Teaching Experience 🐯

Invitied Speaker

• Apr. 2025 Commonsense Reasoning or Hallucination? Unpacking the Plausible Errors of Language Models
  

Invitied Speaker

• Jun. 2023 Introduction to Natural Language Processing and Commonsense Reasoning at Dongguk University
  

Korea University

• Mar - Jun. 2023 TA for NLP for Artificial Intelligence (DFE615-00)
  
• Sep - Dec. 2022 TA for NLP for digital finance engineering (DFE601-00)
  
• Mar - Jun. 2022 TA for Introduction to information retrieval in big data (BDC103-00)
  
• Sep - Dec. 2021 RA for NLP for digital finance engineering (DFE601-00)
  
• Mar - Jun. 2021 TA for Introduction to information retrieval in big data (BDC103-00)
  

NongHyup Bank Co., Ltd

• Jul - Sep. 2020 Teaching data science for new employees

DataDiving

• Jul - Dec. 2020 Invited Instructor
  
• Mar - Jun. 2022 Lecture Producer

Academic Services 😺

Workshop Organizer

• ICLR 2025 Workshop (SSNLP 2025) with A*STAR (Singapore)

Reviewer

• AAAI 2025
• EMNLP 2025
• ICLR 2025
• ACL 2025
• NAACL 2025
• EMNLP 2024
• ACL 2024
• NAACL 2024
• EMNLP 2023
• ACL 2023

Selected Projects 🌌

National Representative AI — World Best LLM

• Research Engineer at Korea University (National Initiative with NC AI, ETRI, KAIST, SNU, LOTTE Innovate, POSCO DX)
• Supported by NIPA, IITP, and NIA
• Built a foundation model from scratch with 1,000 NVIDIA H100 GPUs
• Contributed to large-scale LLM post-training, reasoning evaluation, Mixture-of-Experts (MoE) modeling, dataset construction, and reinforcement learning

Benchmark Dataset across Eight Asian Countries

• Project Manager at Korea University (Collaborative Research with Singapore A* Research)
• Led the design and evaluation of a multi-modal East-Asian benchmark for LLMs
• Coordinated collaboration with research teams across eight countries: Singapore, South Korea, China, Japan, Mongolia, Vietnam, Indonesia, and India

Unlearning for LLM Reliability

• Project Manager & Research Engineer at Korea University (Industry–Academic Joint Project with KT)
• Developed knowledge editing methods tailored for vertical LLMs
• Proposed RAG-based editing approaches with lifelong learning capabilities
• KoLEG: On-the-Fly Korean Legal Knowledge Editing with Continuous Retrieval

Next-Generation AI Technology Development — Hallucination

• Project Manager & Research Engineer at Korea University (National Initiative, with KAIST, SKKU, KU, KETI, NCAI, O2O)
• Supported by IITP
• Secured research funding of 4.8B KRW (≈ $3.5M) over 3 years
• Led research on hallucination detection, mitigation, and domain adaptation
• Best Paper Award: Post-negation Text Induces New Hallucinations in Large Language Models
• Published at ICLR 2025: OpenReview Link

Large Language Modeling for Special Technical Documents

• Project Manager & Research Engineer at Korea University (Industry–Academic Joint Project with GAON)
• Proposed a RAG-based LLM for predictive maintenance in nuclear power plants
• Developed an on-premise deployable model system for handling classified government documents
• Published at EMNLP 2024 (Industry Track): ACL Anthology Link

Open Ko-LLM Leaderboard

• Research Engineer at Korea University (Collaborative Research with Upstage)
• Constructed and deployed the H5 Leaderboard to evaluate commonsense reasoning in LLMs
• Contributed to the evaluation of 2,000+ open-source LLMs
• Developed KoCommonGEN v2, a benchmark for Korean commonsense reasoning challenges
• Published at ACL Findings 2024: ACL Anthology Link

UI/UX Artificial Intelligence Service Development

• Research Engineer at Korea University (Industry–Academic Joint Project with Samsung Fire & Marine Insurance)
• Developed an AI-based application service for dementia prevention (Synerpeace)
• Implemented memory-enhancement learning via API calls and developed an automatic diary summarization feature

R&D Voucher: AI Contact Center

• Research Engineer at Korea University (Industry–Academic Joint Project with O2O)
• Developed an AI-based system for evaluating call quality and providing job training for customer service agents

R&D Voucher: AI Interviewing

• Research Engineer at Korea University (Industry–Academic Joint Project with Withmind)
• Developed an AI-powered interview automation service and established an evaluation system

Data Augmentation with Language Generation Research

• Project Manager & Research Engineer at Korea University (Industry–Academic Joint Project with Samsung Research)
• Proposed a data augmentation method using language generation to improve training of a phone-scam classifier

Korean Prompt-based Engineering for Few-shot Learning

• Project Manager & Research Engineer at Korea University (Industry–Academic Joint Project with Hyundai Motor Group)
• Proposed the Plain Template Insertion method for Korean prompt-based few-shot learning
• Developed an instance classifier for low-resource scenarios
• Published at IEEE Access (SCIE): IEEE Xplore Link

Return on Advertising Spend (ROAS) Prediction Research

• Research Engineer at Korea University (Industry–Academic Joint Project with BizSpring)
• Conducted research on predicting ROAS to optimize advertising effectiveness
• Project supported by the Data Voucher Program
• Published at Mathematics (SCIE): MDPI Link

Hybrid Retriever System for Industrial Frequently Asked Questions Research

• Project Manager & Research Engineer at Korea University (Industry–Academic Joint Project with O2O)
• Proposed a hybrid information retrieval method for customer inquiries by combining dense and sparse embeddings
• Introduced scaling and weighting functions to improve robustness to domain-specific terminology
• Published at Mathematics (SCIE): MDPI Link

Korean Multi-turn Response Chatbot Research

• Research Engineer at Korea University (Industry–Academic Joint Project with Samsung SDS)
• Developed a Korean multi-turn chatbot based on BlenderBot
• Published at KCI 2022: DOI Link

Improving the Understanding of Korean-based Generative Models

• Research Engineer at Korea University (Industry–Academic Joint Project with NSR)
• Project supported by the National Security Research Institute (NSR)
• Developed a GPT-based model to improve classification performance

may u needs ✨

• seojae777@korea.ac.kr
• github.com/J-Seo