Tommaso Dolci
PreDoc Researcher @TU Wien
About Me
I am a PreDoc Researcher in the DMKI Lab at TU Wien and a Marie Skłodowska-Curie doctoral fellow within the ARMADA project. My current work focuses on LLM hallucinations, factuality evaluation, and knowledge graph grounding. I am an advocate of Digital Humanism, the initiative to preserve human agency in digital technology.
Learn MoreRecent News
Starting my secondment at the University of Verona (Italy) as part of the ARMADA project, working with Matteo Lissandrini.
Our paper Value-Preserving Architectures for Agentic AI Systems has been accepted at AgenticDev 2026, co-located with the International Conference on Automated Software Engineering. The paper investigates how different multi-agentic architectures support human-centered values such as privacy, fairness, and pluralism (PDF).
New preprint available! We just completed our survey on reducing LLM hallucinations with knowledge graphs, including a comprehensive review of the literature, an improved taxonomy of existing methods, and a discussion of open challenges (PDF).
Recent Papers
Value-Preserving Architectures for Agentic AI Systems
Alessandro Pesare, Tommaso Dolci, Katja Hose, Emanuel Sallinger
41st IEEE/ACM International Conference on Automated Software Engineering (ASE), 2026. To appear.
Reducing LLM Hallucinations with Knowledge Graphs: A Survey
Tommaso Dolci, Milos Jovanovik, Katja Hose
Preprint under review, 2026.
Towards LLM-KG Symbiosis for Reducing Factual Hallucinations
Tommaso Dolci, Milos Jovanovik, Katja Hose
Proceedings of the Workshops of the EDBT/ICDT Joint Conference, 2026, vol. 4192.