Laura Balbi, a PhD student at LASIGE, received the Best Doctoral Symposium Paper Award at the PhD Symposium of this year’s edition of the Extended Semantic Web Conference (ESWC 2025), for the paper “Neuro-Symbolic AI for conflict-aware learning over Knowledge Graphs”. The conference took place in Portoroz, Slovenia, from the 31st of May to the 6th of June.

The paper addresses the challenges involved in constructing and updating Knowledge Graphs (KGs), which often rely on merging heterogeneous data sources that can introduce ambiguities and contradictions. Current methods for detecting contradictions typically resolve conflicts by discarding one of the contradictory statements, often ignoring the potential for a nuanced, composite truth. Laura’s research investigates how a neuro-symbolic approach — integrating symbolic and Language Models neural representations — can enhance contradiction detection and improve KG representation learning robustness, reliability, and predictive performance, paving the way for more trustworthy machine learning applications over KGs.

The paper can be found here.