Jae-Yon Lee이재연 · UNIST

Research

My research combines Korean literary studies, media history, digital humanities, computational methods, and generative AI. I am especially interested in how quantitative and computational approaches can reshape our understanding of modern Korean literature and its institutions.

Current Research Projects

Stylistic Classification and Generation of Science Fiction in Poetic Prose

This project designs an AI-driven framework that projects Han Kang's dense, interior-focused poetic prose onto a science fiction imagination to generate long-form speculative fiction. The central question asks how we can computationally define and classify a writer's nuanced literary style so that AI-generated text can meaningfully reflect a distinctive literary voice. The project develops an inductive methodology that derives an empirical taxonomy of poetic prose from literary criticism and uses it as both a style-transfer pipeline and an evaluation rubric.

Related Paper
“From Poetic Prose to Science Fiction: Classifying and Generating Han Kang's Literary Style via Computational Methods,” ADHO Digital Humanities Conference (DH2026), Daejeon Convention Center, 2026.07.29 [Forthcoming]

Rewriting Classical Stories as Modern Fiction: Through Changes in Character and Plot

This project introduces a computational framework for rewriting classical literary stories into modern fiction by extracting and transferring structured narrative profiles. It asks how an LLM pipeline can overwrite characters and plots into a new narrative skeleton while automatically detecting and resolving structural or consistency conflicts. The project uses a 3×3×3 matrix that decomposes narratives into 27 cells of character-event contexts, offering a scalable and interpretable computational scaffold for cross-tradition literary adaptation.

Related Paper
“Character Overwriting with LLMs: A 27-Cell Representation for Comparing Literary Umwelts,” Japanese Association for Digital Humanities (JADH) Annual Conference, Kyushu University, 2026.09.12 [Forthcoming]

Colonial and Post-Colonial Sensations and Emotions: A Study of Yi Sang and Yŏm Sang-sŏp

This project uses computational stylistics to investigate the representation of bodily sensations and emotions in the fiction of Yi Sang and Yŏm Sang-sŏp. It asks how we can computationally classify the language of the body and mind to map the stylistic, aesthetic, and affective politics of modernist and realist literature under colonial rule. The project combines a customized multi-level sensory classification model with deep-learning emotion analysis on a digitized corpus of 68 literary works.

Related Paper
“A Quantitative Approach to Bodily Sensations: Modernist and Realist Authors in Colonial Korea,” ADHO Digital Humanities Conference (DH2025), Universidade NOVA de Lisboa, Lisbon, Portugal, 2025.07.17
Built with v0