Research / Emotion
Emotion
The sentiment analysis methodology line at the intersection of AI and the literary humanities. Greybox XAI for diachronic sentiment, the translation work on Mistral and emotional nuance, and the affective dimension of narrative analysis.
Methodology
- The Shapes of Stories (cross-listed with Storytelling) — ensemble methods for diachronic sentiment analysis applied to narrative emotion.
- “Greybox XAI for Diachronic Sentiment Analysis” (Chun and Elkins, International Journal of Digital Humanities 5.2, 2023) — develops the Ensemble Cross-Correlation (ECC) and Ensemble Polarity Confidence (EPC) metrics plus a human-in-the-loop XAI framework. Engaged by Cugurullo and Xu in Policy and Society 44.1 (2024) as the theoretical authority on LLMs' obscure epistemological process — cross-disciplinary travel into political theory of urban governance.
Translation and emotional nuance
- “In Search of a Translator: A Computational Investigation of Proust's Du côté de chez Swann” (Elkins, Frontiers in Computer Science 6, 2024) — Lydia Davis emerges as best literary translator; Mistral best captures emotional nuance among LLMs tested. Engaged by Huang and Cheung in Humanities and Social Sciences Communications (Nature portfolio, 2026). Cross-listed with Language.
Theoretical work on AI and emotion
- “What the Rise of AI Means for Narrative Studies: A Response to Angus Fletcher” (Chun and Elkins, Narrative, 2022).