Keynote
Keynote Speaker Francesca Tomasi - Do Semantic Web technologies still matter?
Francesca Tomasi
University of Bologna
Do Semantic Web technologies still matter?
In the current landscape shaped by the rapid diffusion of Large Language Models (LLMs), the role of Semantic Web methodologies and technologies deserves renewed attention. While LLMs demonstrate remarkable capabilities in processing and generating natural language, they also highlight the importance of structured and semantically explicit knowledge infrastructures capable of ensuring transparency, reliability, and traceability of information. Within the cultural heritage domain, Semantic Web approaches provide a framework for enriching descriptive metadata through the reuse of shared conceptual models and ontologies. Models as “Digital hemeneutics” (Daquino, Pasqual, Tomasi 2020) allow institutions to represent not only the factual description of objects but also the processes through which knowledge about those objects has been produced, interpreted, and revised over time. Traditional archival and bibliographic descriptions, but also digital scholarly editions and full-text document collections, generally focus on recording objective information: authorship, dates, places, or institutional contexts. However, this descriptive layer often omits the interpretative dimension that underlies many of these assertions. The transition from traditional description to semantic models requires the explicit representation of contextual and interpretative information in order to produce reliable and trustworthy data.
In this perspective, RDF vocabularies should not only describe objects but also document the intellectual processes through which knowledge about those objects is constructed. This is particularly important in cases characterized by uncertainty or scholarly debate, such as disputed authorship, uncertain dating, or conflicting historical interpretations. Knowledge graphs offer a powerful solution for representing these interpretative layers. Instead of describing a resource through a single record, a knowledge graph models it as a network of entities, events, and assertions, each associated with its provenance and scholarly context. This narrative dimension is particularly relevant in the era of AI-driven knowledge systems. As LLMs increasingly rely on structured data sources to support reasoning and retrieval, the availability of transparent, provenance-aware knowledge graphs becomes essential. Enriching, both manually and automatically, metadata with interpretative and contextual information not only increases users’ trust but also strengthens the epistemic reliability of cultural heritage data infrastructures. Ultimately, Semantic Web technologies provide the conceptual and technical framework necessary to represent the dynamic and interpretative nature of cultural heritage knowledge. By combining ontologies, provenance models, and knowledge graphs, archives and libraries can move beyond static descriptive records and toward richer representations that capture both the history of objects and the history of their interpretations.