New Research Contribution by Indeewari Balasooriya, Anna Dalla Vecchia, and Elisa Quintarelli, University of Verona.

The paper introduces a hybrid framework combining Virtual Knowledge Graphs (VKGs) with LLMs to enable natural language querying over heterogeneous data sources. Factual questions are translated into SPARQL via NL2SPARQL, while analytical questions, such as tourist preference inference, are routed to a model based inference module. The approach was validated on a real world tourism case study in Verona, Italy.

This work contributes to making knowledge graphs more accessible to non expert users through natural language interfaces.