Context and background
Achieving knowledge interoperability — the integration of independently developed ontologies, knowledge graphs, and databases — is a central challenge in the Semantic Web. These knowledge sources implicitly reflect the perspectives of their creators, leading to conflicts when merged. For example, one hospital may model a tumour as a cellular process, while a laboratory annotates it as a lump of tissue. Naively merging these views yields an inconsistency that makes the combined knowledge base unusable for automated reasoning.
Standpoint Logic (SL) is a multi-modal logic, developed by the project coordinator, that represents knowledge relative to diverse, possibly conflicting standpoints using modal operators. This allows heterogeneous sources to be safely integrated without sacrificing consistency. Recent results show that standpoint extensions of widely-used description logics such as EL and SHIQ preserve their original computational complexity, making the framework both expressive and attractive for applications [1–4].
Objectives
The main objective of this thesis is to establish efficient algorithms and develop reasoning support for Standpoint Logic and apply those to solving knowledge integration problems in the Semantic Web. Concretely, the work will include:
- OWL reasoning via DL translations: design and implement equisatisfiable translations from standpoint description logics (Standpoint-SHIQ, Standpoint-SROIQ) to their base DL, enabling the use of existing optimised OWL reasoners.
- Dedicated algorithms for standpoint reasoning: Establish algorithms based, for instance, on tableau methods or Datalog calculus. Implement and optimise reasoners based on those, either on top of high-performance Datalog engines or extending existing DL reasoners.
- Standpoint-aware ontology alignment: investigate how can SL support richer, multi-perspective alignments between ontologies and entities, going beyond the single unified output of classical matching. Results will be evaluated using synthetic data and/or benchmarks such as those from the Ontology Alignment Evaluation Initiative (OAEI).
