Jeongwan Shin

About Me

Jeongwan is a researcher in the field of Natural Language Processing (NLP), with a focus on machine learning, natural language processing, abstractive summarization, fact correction and Large Language Model (LLM). He received his Ph.D. in computer science and engineering and is interested in forming an LLM-based ecosystem. Recently, as a postdoctoral researcher, he has been studying the integration of radar sensing with natural language and further aims to develop LLMs that enable autonomous systems to interact with the physical world through natural language.

Research Interests and Skills

Natural Language Processing: Language Generation, In-context learning, Summarization, Text-Diffusion generation, Rainforment learning from human feedback
Visual Language Processing: Video Question Ansering
Machine Learning: Generative Model, Representation learning
Radar & Autonomous Systems: mmWave Radar Processing, Radar-Language Alignment, Multimodal Reasoning for Autonomous Systems
Programming: Pytorch, Pytorch lightning
GPU server Build Management: Linux configuration and engineering, NAS, Docker
Language Skills: Korean, English

News

[Nov. 2025] Our paper, "ReFineVQA: Iterative Refinement of Video Description via Feedback Generation for Video Question Answering", has been accepted to WACV 2026.
[Aug. 2025] Started a Postdoctoral Researcher position at DGIST (KAIST InnoCORE LLM Innovation Research Center), focusing on radar–language integration and autonomous systems.
[May. 2025] Excited to share that our paper, "Cross-Modal Knowledge Distillation for Enhancing Domain-Specific Retrievers", has been accepted for presentation at the *Information Retrieval's Role in RAG Systems* (IR-RAG) at SIGIR 2025.
[Mar. 2025] Our paper, "Does Video Language Models Really Understand the Video Contexts?", has been accepted to NAACL 2025 Student Research Workshop (SRW).
[Sep. 2024] Our paper about Factuality Guided Diffusion-Based Abstractive Summarization is accepted to IEEE access 2024.
[Jan. 2023] Our paper about Token-Level Fact Correction in Abstractive Summarization is accepted to IEEE access 2023.

Publications

International Conference

WACV26

ReFineVQA: Iterative Refinement of Video Description via Feedback Generation for Video Question Answering

Jeongwan Shin, Chan Hur, Seongmin Cho, Jae-Ho Choi, Hyeyoung Park

The IEEE/CVF Winter Conference on Applications of Computer Vision 2026, March 6-10, 2026

PDF WACV
SIGIR

Cross-Modal Knowledge Distillation for Enhancing Domain-Specific Retrievers

Jinhyeong Lim, Jeongwan Shin

The 48th International ACM SIGIR Conference on Research and Development in Information Retrieval July 13-17, 2025

PDF SIGIR
NAACL

Do Video Language Models really understand the video contexts?

Jeongwan Shin, Jinhyeong Lim, Hyeyoung Park

The 2025 Annual Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics (NAACL 2025) will be held in Albuquerque, New Mexico from April 29–May 4, 2025.

PDF NAACL

International Journal

Applied Computing eJournal

FreeHal: A Framework for Editing Large Language Model Hallucinations

Jinhyeong Lim, Jeongwan Shin, Yonghwan Lee, Hyun-Je Song

Applied Computing eJournal, 2024.

PDF Site Applied Computing eJournal
IEEE access

Factuality Guided Diffusion-Based Abstractive Summarization

Jeongwan Shin Hyeyoung Park, Hyun-Je Song

Published in: IEEE access( Volume: 12), 2024.

PDF Page SCIE
IEEE access

Token-Level Fact Correction in Abstractive Summarization

Jeongwan Shin, Seong-Bae Park, Hyun-Je Song

Published in: IEEE access( Volume: 11), 2023.

PDF Site SCIE

Domestic journal

Journal of KIISE

Solving Factual Inconsistency in Abstractive Summarization using Named Entity Fact Discrimination

Jeongwan Shin, Yunseok Noh, Hyun-Je Song, Seyoung Park

Korean Institute of Information Scientists and Engineers, vol.49, no.3, pp. 231-240, 2022.

PDF Site KCI

Domestic conference

Atomic Unit-based Post Editing for Hallucination Reduction

Yonghwan Lee, Jeongwan Shin, Hyun-Je Song
The 35th Annual Conference on Human Cognitive Language Technology, 2023.

Multi-view Spectral Clustering for news articles

Sang-Seok Park, Jeongwan Shin, Seyoung Park, Seong-Bae Park
Korean Institute of Information Scientists and Engineers, Korea Computer Congress 2018

Generating Multi-Class Images Using a Generative Adversarial Network

Jeongwan Shin, Yunseok Noh, Se-young Park
Korean Institute of Information Scientists and Engineers, Korea Computer Congress 2017

Neural Conversational Model for Specific Domain

Sujeong Choi, Jeongwan Shin, Ian Jung, Se-Young Park, Seong-Bae Park
Korean Institute of Information Scientists and Engineers, Korea Software Congress 2017

Pattern Filtering for Knowledge Extraction Using Facts from Web

Sol Kim, Jeongwan Shin, Se-Young Park, Sang-Jo Lee
Korean Institute of Information Scientists and Engineers. 2015

Experience

Invited Talk

Understanding the Objective Function of RLHF(From ChatGPT to Deepseek)

PLACE: Kyungpook National University(2025)

PDF
Practical Testing of RAG for PDF Question Answering

PLACE: Jeonbuk National University(2024)

PDF
Table structure data correction

PLACE: Jeonbuk National University(2024)

PDF
Diffusion Model for Text Generation

PLACE: Kyung Hee University(2023)

PDF
Token-Level Fact Correction in Abstractive Summarization

PLACE: Jeonbuk National University(2022)

Internship

KAIST Hangul Engineering Research Lab (Mar 2013 - Feb 2014)

Development of Hangul-based Universal Phonetic System
Developing keyboard applications for iPhone and Android
Technology integration and call process development