| Description | Knowledge graphs will play a central role in the forthcoming Artificial Intelligence (AI) landscape. They enable the representation of structured, machine-readable knowledge and provide the backbone for combining neuro-symbolic and symbolic approaches, i.e., merging statistical learning with logical reasoning. This integration is increasingly recognized as the future of AI, as it allows AI systems to move beyond purely statistical correlations and instead operate with explicit, interpretable domain knowledge.
In parallel, Semantic Web technologies, particularly RDF, are being adopted more widely across both public institutions and private enterprises. Ensuring that AI systems remain compatible with these standards is essential for interoperability, scalability, and seamless integration into existing infrastructures. This alignment is critical not only for embedding domain expertise into real-world applications but also for ensuring that innovation leads to usable, practical solutions.
Yet, despite their promise, much work remains to make knowledge graphs truly usable in practice. A major milestone in this direction has been the introduction of SHACL, the W3C recommendation for validating RDF knowledge graphs (https://www.w3.org/TR/shacl). SHACL provides a standardized way to check the consistency and quality of RDF data, significantly advancing the reliability of KGs.
SHACL is now evolving to support inferencing over RDF data (https://www.w3.org/2024/12/data-shapes.html), while new initiatives such as SHACL-X merge SHACL with JavaScript, extending its expressive power to that of a general-purpose programming language. However, so far, these innovations have been tested mostly on small, artificial examples. The next step is to apply them to the design and development of full ontologies that can drive real-world AI applications.
This project will pioneer the creation of such an ontology. The specific focus of the ontology is open to discussion. Please contact Livio Robaldo (livio.robaldo@swansea.ac.uk) to agree on the scope. Possible directions include:
- An ontology to represent obligations, permissions, and other deontic concepts.
- An ontology to represent temporal information.
- An ontology to represent spatial information.
- An ontology to represent natural language quantifiers.
- Etc.
By the end of the project, the student will be familiar with the latest advancements in the Semantic Web and will gain hands-on experience with the key technologies driving the next generation of AI services. This expertise will be especially valuable as AI increasingly converges on the integration of neuro-symbolic and symbolic knowledge. |
|---|