Masterarbeit

Scientific knowledge graphs require continuous maintenance as publications, datasets, repositories and software projects evolve. This thesis investigates an agentic approach in which specialized agents detect changes in source resources, determine which KG statements are affected, and propose or perform corresponding updates. Agents will use provenance information to identify the sources and evidence underlying existing statements before modifying the graph. The research will evaluate whether agentic maintenance can improve the timeliness and consistency of a scientific KG while reducing manual curation effort.