Named Entity Recognition in Historical Italian: the Case of Leopardi's Zibaldone

This presentation, delivered at the X-Tail Workshop co-located with EKAW 2024, showcases our team's work at the University of Macerata on advancing Information Extraction (IE) techniques for Italian humanistic texts. We introduce a new challenging benchmark for Named Entity Recognition (NER) and Entity Linking (EL), derived from Leopardi’s Zibaldone, a seminal work in Italian literature known for its linguistic complexity and philosophical depth. Our benchmark is specifically designed to address the unique challenges of analyzing historical and humanistic texts, including archaic language, complex syntactic structures, and the diversity of entity types. Additionally, the presentation includes the evaluation of a domain-specific NER model, tailored for humanistic and literary corpora. We explore how such models can enhance the extraction and linking of meaningful entities, providing tools to facilitate deeper textual analysis for scholars in digital humanities. This work bridges the gap between cutting-edge Natural Language Processing (NLP) technologies and the needs of the digital humanities community, laying the groundwork for improved computational analysis of cultural heritage texts.