About the workshop
The recent growth of Large Language Models (LLMs) has expanded possibilities in data management, enabling powerful natural language access, reasoning, and decision support. However, reliability and trustworthiness remain major challenges when deploying LLMs in sensitive domains. Graph-based representations of knowledge and data (e.g., knowledge graphs and property graphs) provide a promising avenue to address these challenges.
LLMs generate fluent responses but often struggle with factuality, bias, hallucinations, and a lack of explainability. Graphs, on the other hand, provide structured, interconnected representations that can serve as grounding and validation layers for LLM-based systems. In this context, Semantic Web technologies offer standardized graph-based frameworks (e.g., RDF and ontologies) to formally model, link, and reason over knowledge.
