COntExt Framework Automates Ontology Extension from Metrics

Organizations increasingly rely on structured operational metrics to monitor systems and processes, but the rich domain knowledge embedded within these metric definitions often remains disconnected from formal ontologies. This manual gap between metric catalogs and ontological knowledge is labor-intensive and inefficient. To address this, researchers have introduced COntExt, a framework designed for context-aware ontology extension. COntExt takes structured metric definitions as input and intelligently suggests how new concepts and properties referenced within these metrics should be integrated into an existing ontology. The framework breaks down the complex extension problem into three sub-tasks: predicting parent classes, identifying relation types, and assigning data properties. It leverages the contextual information derived from the metrics themselves to inform these suggestions. Evaluations across four cybersecurity ontologies demonstrated that metric-derived context significantly enhances the accuracy of suggestions for relation type prediction and data property assignment, outperforming baseline methods that only use ontology context. This work highlights operational metric catalogs as a valuable, yet underexploited, source for maintaining and expanding ontologies at a substantially reduced cost compared to traditional manual engineering efforts.