Exploring multimodal perception in large language models through perceptual strength ratings
IEEE Access, 13, 176751–176769.
Dojun Park
Computational linguist
Hi, I'm Dojun. I study how language models interpret meaning, infer intentions, and use context across languages. My work focuses on computational pragmatics, multilinguality, and LLM evaluation. I aim to connect these insights to the development of more capable language models, bringing linguistic analysis into model development and evaluation.
Publications
Evaluating Large language models on Understanding Korean Indirect Speech Acts
arXiv:2502.10995
Pragmatic competence evaluation of large language models for the Korean language
38th Pacific Asia Conference on Language, Information and Computation · December 2024.
MultiPragEval: Multilingual pragmatic evaluation of large language models
2nd GenBench Workshop on Generalisation (Benchmarking) in NLP, 96–119 · November 2024.
Multi-dimensional machine translation evaluation: Model evaluation and resource for Korean
LREC-COLING 2024, 11723–11744 · May 2024.
Grammar accuracy evaluation (GAE): Quantifiable quantitative evaluation of machine translation models
Journal of KIISE, 49(7), 514–520.
German Phoneme Recognition with Text-to-Phoneme Data Augmentation
arXiv:2211.13776
Korean-English machine translation with multiple tokenization strategy
Korea Computer Congress 2021, 1720–1722 · June 2021.
Background
Education
2021–2023
University of Stuttgart
M.Sc. in Computational Linguistics
Thesis: Multi-score neural machine translation evaluation: beyond single score metrics
Supervisor: Sebastian Padó
2018–2021
Konkuk University
B.Tech. Convergence, Human ICT
B.A. in English Language and Literature
Thesis: Korean-English machine translation with multiple tokenization strategy
Supervisor: Harksoo Kim
Experience
Dec 2024–Mar 2025
Rivetta
AI Engineer · Seoul, South Korea
Developed an LLM-based Korean–English web novel translation pipeline, including narrative segmentation, automated glossary generation, and cross-system translation quality review.
Nov 2023–Aug 2024
Seoul National University
Research Assistant · AI Institute
Worked on MultiPragEval across English, German, Korean, and Chinese; Korean implicature and indirect speech act evaluation; and perceptual grounding in GPT-4 models across six sensory modalities.
Supervisor: Sungeun Lee
Jul–Dec 2020
Konkuk University NLP Lab
Undergraduate Researcher
Trained Transformer-based Korean–English translation models with different tokenization strategies. Proposed Grammar Accuracy Evaluation (GAE) to complement BLEU and compared the metrics across four translation directions.
Supervisor: Harksoo Kim
Academic service & languages
Reviewer
- 2026Journal of Pragmatics
- 2025Korean Journal of Textlinguistics
- 2024Spanish Journal of Applied Linguistics
Languages
- Korean
- Native
- English
- TOEFL 100
- German
- DSH 2 · CEFR C1
- Chinese
- HSK 2 · CEFR A2