My research spans digital humanities, computational literary studies, policy analysis, and the application of large language models to humanities research.
In literary studies, I examine war, gender, emotion, and representation across Japanese, British, and Russian literature. My research on Japanese atomic bomb literature uses sentiment analysis and computer-assisted close reading to investigate differences in masculine and feminine emotional modes of writing. I have also studied wartime short fiction by Mollie Panter-Downes, focusing on narrative strategies that reduce or obscure the emotional presence of war. Other projects have addressed Arctic naval fiction, Japanese classical poetry, and the reception of Russian literature in Japan.
In policy studies, I work on data-driven research concerning Arctic politics and public discourse. My projects analyze Japanese parliamentary debates, policy networks, expert commentary, and media discourse to trace the formation of Japan’s Arctic identity and changing attitudes toward Arctic sea routes, governance, and international cooperation. I also collaborate on research into Russia’s post-2022 Arctic discourse and its growing emphasis on cooperation with Asian countries. In these projects, I contribute expertise in data collection, corpus construction, statistical analysis, visualization, and multilingual text analysis.
My recent research explores the use of large language models and generative AI in computational literary studies. I investigate how LLMs can be fine-tuned and customized for literary generation, stylistic analysis, sentiment classification, and other humanities-oriented NLP tasks. This includes experiments with parameter-efficient fine-tuning, literary corpora, and the ability of language models to reproduce or generalize stylistic features.
Methodologically, my work combines digital humanities approaches with natural language processing, corpus linguistics, statistical modelling, and close reading. I develop multilingual research pipelines covering data collection, cleaning, annotation, model inference, evaluation, and visualization. My aim is to use computational methods not as a replacement for interpretation, but as a way to identify patterns, test arguments, and support historically and culturally grounded humanities research.