Research Interest
I am a Ph.D. student at Seoul National University, advised by Prof. Seungwon Hwang. I want to develop NLP systems that are beneficial and informative to people in everyday life. To achieve this, I am focusing on endowing language models with the ability to:
(1) Efficiently stay up-to-date with new knowledge and human feedback,
(2) Incorporate internal (commonsense) knowledge with external knowledge,
(3) Generate reliable responses grounded on external knowledge base.
Publications
- Soyoung Yoon*, Jongyoon Kim*, Seung-won Hwang. (2024) Analyzing the Effectiveness of Listwise Reranking with Positional Invariance on Temporal Generalizability., CLEF 2024, LongEval. paper competition results. (ranked #8th)
- Soyoung Yoon, Eunbi Choi, Jiyeon Kim, Hyeongu Yun, Yireun Kim, Seung-won Hwang. (2023) ListT5: Listwise Reranking with Fusion-in-Decoder Improves Zero-shot Retrieval., ACL 2024 (main, oral - acceptance rate 102/4407=2.3%) paper | code | poster | slides | presentation(video)
- Jinkyung Jo, Dayeon Ki, Soyoung Yoon, Minjoon Seo. (2023) An Integrated Search System for Korea Weather Data, EMNLP 2023 Industry Track. paper
- Chaeeun Kim*, Soyoung Yoon*, Hyunji Lee, Joel Jang, Sohee Yang, Minjoon Seo. (2023) Exploring the Practicality of Generative Retrieval on Dynamic Corpora, EMNLP 2024, also appeared on GenIR@SIGIR 2024. paper
- Soyoung Yoon, Sungjoon Park, Gyuwan Kim, Junhee Cho, Kihyo Park, Gyu Tae Kim, Minjoon Seo, Alice Oh. (2022) Towards standardizing Korean Grammatical Error Correction: Datasets and Annotation, ACL 2023 (main) paper | Code&Data | Poster | Demo
- Soyoung Yoon*, Gyuwan Kim*, Kyumin Park. (2021) SSMix: Saliency-based Span Mixup for Text Classification, ACL 2021 (Findings), also appeared on the 6th Workshop on Representation Learning for NLP(Rep4NLP) paper | code | blog (* Equal Contribution)
Education
- Seoul National University, PhD., Language & Data Intelligence Lab, Aug 2023 - Present (advised by Seung-won Hwang) GPA 4.0/4.3
- KAIST, M.S., Language & Knowledge Lab, Graduate school of AI, Mar 2021 - Aug 2023 (advised by Minjoon Seo) GPA 4.1/4.3
- KAIST, B.S., Computer Science, Mar 2016 – Feb 2021 Advanced major on Computer Science. Completed courses focused on AI. GPA 3.77/4.3(Cum Laude). Major only: 3.92/4.3
Work Experience
- Research Intern, Channel.io, AI team, (Jan 2024 - Jul 2024) -> Working on input table reformulation for better RAG
- Research Intern, Exaone Lab, LG AI, (May 2023 - Nov 2023) -> ACL 2024 ListT5 paper
- Research Intern, Hyundai AIRS, (Now 42dot.ai), (Dec 2021 - Feb 2021) -> Worked on text-to-SQL semantic parsing
- Research Intern, Clova AI, Naver Corp, (July 2020 - Jan 2021) Advisor: Gyuwan Kim. -> ACL 2021 SSMix paper
- Research Intern, KAIST, U&I Lab (Aug 2019 – July 2020) Advisor: Alice Oh, Sungjoon Park. -> ACL 2023 GEC paper
- SW Developer Intern, Aitrics, (Jan 2019 – Aug 2019) -> talk at pycon
Awards
- National Graduate Science & Technology Scholarship, Mar 2018 - Present
- 3rd prize, URP workshop(Undergraduate Research Project), Sep 2020 (Description in Korean / Workshop presentation slides / List of winners)
Talks / Blog posts / Community contributions
- 2024.7.8 Published a tech blog post about "Table QA with LLMs - Huge table parser by pseudocode filtering" on the channel.io (internship company) tech insight blog. blog link
- 2023.10.5 Huggingface contributor - Pull request to fix the t5x conversion script merged at huggingface main
- 2020.6.9 KENS(KAIST Educational Network System) contributor - KENS is an official project for KAIST CS341: Introduction to Computer Networks, which is a 4-credit project-oriented course that KAIST CS undergraduates take.
- 2019.8.14 Django Query Optimization for Real-time medical AI data processing, Pycon Korea. (Slides / Presentation)
Projects (Academic)
- Classification and summarization model of parliamentary members' speeches based on the National Assembly minutes
- proposal | presentation | poster
- Which Model is Helpful in Solving Privacy, Memorization, and Bias Problems?
- proposal(presentation) | progress report(presentation) | final(presentation) | paper
- Academic Paper Writing Tone Depending on the Author's Location
- proposal(presentation) | proposal(video) | final presentation | paper
- Debiasing Youtube Comments by clustering based on perspectives
- project pitch | milestone | final | code | code(backend) | paper
- Multi-modal Movie Box Office Prediction
- presentation | code | paper
- Extracting Character Information from Movie Script
- code | proposal(paper) | interim(paper) | final(paper) | presentation
- Trend Analysis and Event Tracking using Topic Modeling
- presentation | paper | code
- Regularizing and Optimizing RNN Language Models
- paper | proposal | presentation
Projects (Funding)
- Korea National Institute of Meteorological Sciences (NIMS), Developing an AI-based forecast support solution. (2021)
- Hyundai Motor Group, Research on effective natural language understanding methods in resource-constrained environments. (2022)
Extracurricular activities & Experiences
- Became a member of CONNECT Translating supporters, 2018 - 2020 (Marked as translation contributor for deep learning courses such as School of AI, and Fast.ai)
- Freshmen Guidance Group(Proctor), Mar 2018 - Dec 2018
- Participated as an instructor on Samsung Dreamclass camp, Jan 2017 - Feb 2017
- KAIST Buddy program, 2017
- Joined KAIST GoN team, 2017 - Present
- Joined KAIST pop band Carpe Diem as a female vocalist, 2016 - Present
- Participated at Education Program for Gifted Youth(EPGY), summer 2010
- Lived at Palo Alto, California, July 2009 - June 2010
Reviewing
- ACL Rolling Review 2023 December (ARR), 2024
- European Chapter of the Association for Computational Linguistics (EACL), 2023
- Automated Knowledge Base Construction (AKBC), 2022
Language Proficiency
- Korean - Native
- English - High. TOEFL: 108/120 (Nov.2018), TOEIC: 975/990 (June.2020)
Personal
- I really love singing (Wastin' / Slow Motion / Speechless)
- Featured, original songs (officially published): Fly High / Trip
- I also like taking pictures - Instagram and do knitting - Ravelry Page
- My energy comes from meeting new people and having new experiences, so feel free to mail \& collaborate with me!
- Come visit my Blog!
Contact
Soyoung Yoon
soyoung.yoon (at) snu.ac.kr
recoprin (at) gmail.com
Ph.D Student
Seoul National University
Seoul, Republic of Korea
(Last Update: 06/06/2024)