Reference Professor: Francesca Tomasi. Doctoral Credits: 6 (30h lectures). Assessment: Pass/Fail. Years: I, II, III
Date:
Event location: In presence and online event
Type: Curricular Class
Training activity that aims to introduce the topic of the life cycle of digital cultural resources. The methods, models and critical perspectives of Digital Humanities will be described for the production, description, management, processing, elaboration and dissemination of heritage.
On June 12, there will be two sessions: a philosophical reflection on LLMs led by Davide Picca, and a lecture by Federico Condello, during which we will be able to discuss philology.
June 19 will be dedicated to exploring resources and tools for Digital Humanities.
Davide Picca, When Words Reflect Many Worlds: A Semiotic-Aware Framework for Evaluating the Fidelity and Coverage of LLM-Generated Outputs (link to participate)
Language plays a fundamental role in human communication and in human-machine interaction. However, the meaning of a linguistic expression is not determined by its wording alone. The same expression can be understood in different ways depending on context, background, and perspective. In this work, we introduce a semiotic-aware framework for evaluating how Large Language Models (LLMs) interpret a target expression. Our method represents what an output expresses about a concept beyond its wording, focusing on the interpretation it activates and on the salient elements connected to that interpretation. These representations are compared through a non-parametric manifold-based approximation, from which we derive two complementary measures, Semiotic Fidelity (SF) and Semiotic Coverage (SC). SF captures whether an LLM remains aligned with human interpretations, while SC captures whether it recovers the breadth of meanings expressed by humans. Experiments show that our measures capture interpretive differences that existing evaluation approaches often overlook. They distinguish semantic shifts from harmless paraphrasing, reveal unsupported interpretations introduced in LLM outputs, and identify cases where only part of the human meaning space is recovered. By making these differences explicit, the framework provides an interpretable way to study meaning in AI systems and to connect computational evaluation with broader theories of language and interpretation.
Lecture by Federico Condello, La filologia, ovvero la lotta per formalizzare l’informe (link to participate)
Federico Condello (1973) is a full professor of Classical Philology at the Alma Mater Studiorum–University of Bologna. He teaches Classical Philology and coordinates the Laboratory for Specialized Translation from Ancient Languages (TraSLA).
As a researcher, he focuses particularly on textual criticism and the history of tradition; Greek epic, lyric, and tragedy; the transmission and reception of the classics; and the theory and practice of translation from ancient languages.
He is a member of the Scientific Committee and coordinates the public activities of the “La permanenza del Classico” Study Center; he is a member of the Scientific Committee of the journal Eikasmós and of several other Italian journals and series; he contributes to the daily newspapers “La Repubblica” and “il manifesto.”
He serves as the University Delegate for Students and the University Delegate for Institutional Communication.
The course will be held during the day and organized as a blended learning program, using the following platforms: DevDH.org, DariahTeach, and Campus, as well as Miriam Posner’s tutorials. (link to participate)
Each student will be able to choose to explore one or more of the PhD program’s topics related to:
1. Computational thinking and data management: concepts, tools, and applications
2. Information modeling and web technologies
3. Information science and knowledge organization in archives and libraries
4. Digital scholarly editing of texts
5. Digital cultural heritage and interactive media
6. Information visualization
7. Designing for the DH
8. Frontiers of artificial intelligence
9. The evolution of paradigms in computational linguistics
1 Doctoral Credit = 15h lectures
1 Doctoral Credit = 5h lectures
1 Doctoral Credit = 5h lectures
1 Doctoral Credit = 10h lectures
Check the PhD Agenda