MASTERÂ Assignment
SPARQL Authorization at Scale: A Comparative Experiment on Large Scale Knowledge Graphs
Type : Master M-BIT
Period: February - July, 2025
Student : Mol, S. (Sven, Student X-BIT)
Date Final project: July 21, 2025
Supervisors:
Abstract:
Knowledge graphs see a growing adoption across a wide variety of domains. A significant barrier to adoption remains, however, the lack of support for fine-grained authorization. While various approaches to enforce authorization on knowledge graphs have been proposed in literature over the years, they are often evaluated on relatively small datasets. This leaves the question of scalability open to discussion. This thesis investigates the applicability of these methods on large scale knowledge graphs. Fundamental flaws are found with inspection and filtering based approaches, while encryption is not suitable to the further problem context. The scalability of query rewriting and data preselection is investigated using different triple stores. This shows that the performance is largely dependent on the underlying triple store. In general query rewriting offers better performance at scale, however, data preselection can be applied when the generated selection can be persisted and reused.