What is this thesis going to cover?
Among the key semantic technologies, there are RDF and OWL, which are used together to represent and manipulate knowledge in the form of ontologies. In particular, RDF is a format for explicit data, similar to how it is done in relational databases. At the same time, OWL is used to represent implicit knowledge, which is usually a set of logical axioms that can be applied to enrich the explicit data. Interestingly enough, the OWL axioms are also written in RDF format, so that the overall RDF knowledge base has heterogeneous nature and consists of two parts.
One of the main obstacles for deployment and usage of RDF and OWL in real life is the difficulties with knowledge base development and maintenance. In essence, an issue is that RDF is extremely fine-grained, and thus it is extremely difficult to manage the large and complex semantic of a knowledge base as a whole.
A promising formalism to overcome this obstacle is the ontology template language OTTR, which has been recently developed by the Analytical Solutions and Reasoning (ASR) group. OTTR templates are similar to basic macros known from programming languages and allow complex ontology structures to be encapsulated as templates and given a clear interface for use. We expect that ontology templates will dramatically reduce the cost and barriers of using semantic technologies and ontologies at scale.
The industry has expressed interest in the OTTR approach and we are working together with several companies, including Aibel and DNV-GL, to further develop tools and methodologies for large-scale industrial use of ontology templates.
In this project, you will work on establishing connections between OTTR and related Semantic Web formalisms, namely, SPARQL and SHACL. SPARQL is a default query language for RDF knowledge bases and resembles SQL, while SHACL is a formalism for logical constraints on the shape of RDF. Specifically, your work will have both a theoretical and a practical component. From the theoretical side, you will identify the fragments of OTTR that are rewritable to SPARQL and SHACL and vice versa, as well as develop theoretically worst-case optimal algorithms for these rewritings. Then, from the practical side, you will implement your algorithms and test their performance on both synthetic and real-life benchmarks.
As explained above, the current primary use of OTTR templates is to create ontologies. The benefit of allowing OTTR templates to be rewritten to SPARQL queries and SHACL shapes is that OTTR templates can then also be used for querying ontologies and for describing constraints over ontologies. One may then be able to build and maintain a library of OTTR templates that caters for many different aspects of building and interacting with ontologies, all using one common framework.
Why should you apply for this thesis?
This project offers an opportunity to get rounded experience in both theory and practice of semantic technologies on the edge of their current development. It will allow you to realise the whole cycle of your own scientific project, starting from an idea, continue by proving theoretical results, and completing with an implementation and evaluation. At the same time, you will be part of an active research and innovation project, and get the chance to evaluate your work ongoing industrial projects.
Who should apply?
We are looking for a highly motivated master student who is genuinely interested in getting into the theory and practice of modern semantic technologies. Having taken IN3060/IN4060 (Semantic Technologies) will be a strong advantage, since an introduction to OTTR templates is given in this course. Besides this, your implementation will build on an existing open source project written in Java, so strong skills in Java and experience with Git is also a benefit.