ResearchPod Summary
In the defence and national security sector, the use of knowledge graphs and ontologies is critical for systematic sensemaking and the development of robust, neurosymbolic AI systems. However, the domain is characterized by extreme breadth and specialization, leading to a fragmented landscape of heterogeneous ontologies developed by academia, government, and industry. Achieving interoperability across these diverse models is a significant challenge that currently hinders the effective integration of data for security applications.
To address this, the authors curated the Defence, Intelligence and Security Ontologies (DISO) collection. This repository aggregates over 60 publicly available OWL ontologies, categorized into 11 clusters including cybersecurity, situation awareness, and smart environments. The authors performed a bulk alignment exercise using the LogMap system across 1,653 ontology pairs, demonstrating significant conceptual overlap and potential for integration across the network.
To foster better interoperability, the authors introduced a new track for the Ontology Alignment Evaluation Initiative (OAEI). This track features eight specific matching tasks selected based on their complexity and relevance to the domain. The authors generated a consensus alignment by aggregating outputs from multiple state-of-the-art matching systems and created a manually curated silver-standard reference. This resource provides a standardized benchmark for researchers to test and improve ontology alignment tools specifically for security-related vocabularies.
By providing a centralized, FAIR-compliant repository and a rigorous evaluation framework, this work enables the development of more reliable, semantically sound AI systems for defence. It bridges the gap between general-purpose upper-level ontologies and highly specialized domain models, facilitating the systematic integration of heterogeneous data sources essential for modern national security operations.
AI-generated third-party summary by ResearchPod. Not official content or an endorsement by the paper authors or affiliated organizations.