Seminario "Semantic Similarity for Humans and Machines" (Cátia Pesquita)

Semantic Similarity for Humans and Machines

Quando: 26 ottobre 2026 – ore 10:00​

Dove: Aula Consiliare DEI​

ABSTRACT
Semantic similarity plays a fundamental role in how humans interpret and relate concepts, and in how machines represent and exploit knowledge. In this talk, I will explore different approaches to measuring semantic similarity, from ontology-based methods grounded in structured domain knowledge to representation learning approaches based on knowledge graph embeddings. Drawing on examples from the biomedical domain, I will discuss the strengths and limitations of these approaches, particularly their ability to capture meaningful semantic relationships. I will also explore how symbolic and subsymbolic representations can complement each other, highlighting opportunities for more interpretable, knowledge-aware AI systems.

BIOGRAPHY

Cátia Pesquita is a Professor In AI and Data Science at the Faculty of Sciences of the University of Lisbon, and Vice-Director of LASIGE, where she leads the Health and Biomedical Informatics research line. Her research explores how to bridge symbolic knowledge and machine learning to build AI systems that are more knowledgeable, interpretable, and capable of supporting scientific discovery. She has made contributions to biomedical semantic similarity, ontology integration, and knowledge graph-based learning, with a particular interest in leveraging structured scientific knowledge to improve AI predictions and explanations. She coordinates the BSc in Data Science at Ciências ULisboa, has authored over 70 scientific publications, and has been recognised among the world' s top 2% most cited scientists since 2020.