The program is under construction. It will be annouced soon.

Chiara Palladino (Furman Univesity)
The fragile link: Semantic Annotation and the study of the past

Abstract
Semantic annotation is one of the most powerful tools of current digital scholarship. With the raise of machine learning and increasingly customizable language models, the ability to design and evaluate semantically classified datasets has become essential to the critical understanding of complex documents from the past. Yet, the study of the ancient world challenges semantic annotation at the very foundations. This talk explores the complexities of classifying and annotating ancient spatial documents, as an example of sources that often defy modern conceptual frameworks. The uncertainty and ambiguity of ancient geographical information, the fluidity of semantic categories over time, and the problematic alignment of ancient and modern lexicon represent some of the key challenges to the creation of semantically classified data from ancient spatial corpora.
Current research gravitates towards project-specific annotations to carry the interpretation and discursive analysis of ancient documents. However, we also need some level of interoperability through disambiguation and linking to digitized lexica, indices, and gazetteers. In fact, the specificity of the digital study of antiquity lies in the very fragile nature of the links we establish: between annotations and authorities, but also between ancient categories and modern classifications. While ontologies provide a framework for semantic annotation, the lack of documented correspondence between modern and ancient categories risks leading to data loss, but also to lack of understanding of what lies outside the standard. We argue for a theory of semantic annotation that uncovers the epistemological implications of this task, embracing fragility and ambiguity, and treating special cases not as outliers, but as tools to emphasize the inconsistencies of our knowledge systems.

Margherita Fantoli (KU Leuven)
Beyond Words: Automating Contextual Understanding in Classical Texts

Abstract
Semantic annotation of ancient texts involves two key steps: identifying meaningful tokens, mostly through their linguistic features, and contextualizing them by linking to (disambiguated) dictionaries, knowledge bases, or by labelling specific roles. This process aligns with the ancient tradition of glossing and commenting on texts, a practice central to classical scholarship. While manual annotation is labour-intensive and disambiguated corpora are rare, can we automate this process to create large-scale semantically annotated resources?
The dream of enhancing textual understanding through annotation dates back to Padre Busa, whose lemmatized concordance of Thomas Aquinas’s works laid the foundation for automated linguistic annotation. Despite the evolution of AI models, the automation of annotation of classical texts remains task-dependent, with semantic annotation—requiring extralinguistic context—posing significant challenges.
This keynote explores the potential and limitations of automating semantic annotation for ancient texts. It reviews advances in Named Entity Recognition (NER) and Linking (NEL), examines how digital resources like gazetteers and encyclopedias capture context, and considers how to integrate this knowledge into Large Language Models (LLMs). By bridging linguistic annotation and semantic enrichment, this talk envisions a future where automation enhances the semantic understanding of ancient texts, while upholding the traditions of classical scholarship.