Research Laboratory on Digital Textualities
  1. Home
  2. Events
  3. 2026-07-27

The Lab at the annual DH Conference in Daejeon

On the occasion of the annual conference of the Alliance of Digital Humanities Organizations (ADHO) in Daejeon, South Korea, two members of the Research Lab on Digital Textualities, Alexia Schneider and William Bouchard, will deliver a series of presentations:

Reclaiming Epistemic Agency in AI-Assisted Research: A Recommender System That Reveals, Rather Than Hides

  • Date: July 30 at 10:00 a.m. (UTC+9)
  • Presenters: Schneider, Alexia; Levy, Pierre; & Vitali-Rosati, Marcello

This presentation introduces a recommender system designed to make AI-assisted research more transparent and exploratory. Integrated into the Isidore search engine, the tool contrasts ontology-based semantic retrieval (IEML) with LLM-augmented search, allowing users to visualize how different retrieval models shape the literature they discover. Combining human–AI collaborative semantic translation, a user study, and an evaluation of LLM performance, the project demonstrates how recommender systems can strengthen researchers' epistemic agency by promoting serendipitous discovery, explainability, and a critical understanding of the mechanisms underlying AI-generated recommendations.

Modelling Literary Variation in the Greek Anthology

  • Date: July 30 at 1:40 p.m. (UTC+9)
  • Presenters: Vitali-Rosati, Marcello; Bouchard, William; Audin, Yann; Forest, Dominic

The IAL (Literary Artificial Intelligence) project investigates whether a literary concept can be formalized computationally by modelling epigrammatic variation in the Greek Anthology. Drawing on a corpus annotated by philologists, the presentation compares traditional natural language processing techniques with BERT-based models to assess their ability to identify patterns of literary rewriting while remaining interpretable for scholarly analysis. Rather than focusing solely on predictive performance, the project examines how explainable computational models can support literary interpretation. Its findings suggest that the gap between computational similarity and philological similarity is itself a valuable result, shedding light on the methodological challenges of literary modelling and explainable AI in the Digital Humanities.

Poster: The Anthologia Graeca Project: 11 Years of Editorialization

  • Date: July 30 at 1:40 p.m. (UTC+9)
  • Presenters: Vitali-Rosati, Marcello; Bouchard, William; Verstraete, Mathilde
This presentation reflects on the eleven-year development of the Anthologia Graeca project, an open-access, collaborative digital edition of the Greek Anthology that reimagines scholarly editing as an open infrastructure for research, teaching, and public engagement. Built around a richly annotated corpus of more than 4,000 epigrams, the project combines interoperable data, open APIs, and collaborative editorial workflows that enable scholars, students, and educators to contribute to the edition. By foregrounding editorialization, small data, and knowledge circulation, it demonstrates how a highly specialized corpus can become a shared laboratory for philology, Digital Humanities, and the development of explainable AI methods.