Stefan Ultes

Papers

Total Citations

H-Index

About

Stefan Ultes is a leading researcher in dialogue systems and conversational AI, with a particular focus on advancing dialogue management through structured knowledge representations. His most notable contribution is the introduction of GraphWOZ, a novel approach that leverages conversational knowledge graphs as the core representation of dialogue state. This work, published in 2022, not only presents an innovative framework for more coherent and context-aware dialogue management but also introduces a valuable new dataset—GraphWOZ—comprising Wizard-of-Oz dialogues where human participants interact with a robot receptionist. By grounding dialogue state in graph structures, Ultes addresses key challenges in tracking complex, multi-turn conversations, enabling systems to maintain richer contextual understanding. His research has garnered attention within the community, with his most-cited paper accumulating citations that underscore its growing influence. Ultes’ work bridges the gap between symbolic reasoning and neural approaches, offering a scalable path toward more robust and interpretable conversational agents. For students and researchers, his contributions provide a foundational framework for building dialogue systems that can handle dynamic, real-world interactions with greater fidelity and adaptability.