Linked Data is a standardized way of publishing interoperable data to the Web, with the goal of making it easier for intelligent applications to access, combine and interpret data from heterogeneous sources. While a large amount of Linked Data is published in an open way, this is not always feasible in real-world scenarios, where sensitive data is involved, that cannot be accessible to all. To facilitate this, a query rewriting may be used to enable an authorization layer on top of the underlying database - typically an RDF triple store. However, this approach comes at a performance cost, in terms of query answering speed, computing resources used, stability, etc.
In this thesis, the student will investigate the overhead of an existing SPARQL rewriting engine, compare it to related approaches from literature, make recommendations as to how it could be improved, and implement one or two of these recommendations to assess its impact. The query rewriting engine in question is SPARQL-parser, an open source microservice which is used in the context of redpencil's semantic.works technology stack. This architecture is completely open-source, and already in use in several production environments. In other words, the conclusions and recommendations of this thesis could have a direct and significant impact on real-world, in-use applications.
Company: redpencil.io
Contacts:
- Academic supervisor: Prof. Christophe Debruyne (ULiège)
- Industry supervisor: Dr. Tom De Nies (< e-mail verwijderd om veiligheidsredenen >)
- Administrative contact: Johan Delauré (< e-mail verwijderd om veiligheidsredenen >)
This topic is an applied research topic, and students will be encouraged to be involved in the potential dissemination of the results.