An ontology without governance is a liability rather than an asset. A knowledge graph that was accurate six months ago but has not been updated since becomes a source of systematically wrong answers as its underlying data moves, its reference terminologies are revised, and its modelling assumptions are overtaken by domain developments. The governance processes that prevent this degradation are not glamorous — they involve committees, workflows, version control, and quality metrics — but they are as critical to the value of the knowledge infrastructure as the technical components that built it.
The Governance Structures Required
Effective ontology governance requires three organisational structures. A stewardship team — typically two to four domain experts and one knowledge engineer — owns day-to-day maintenance: processing concept requests, validating incoming data, monitoring quality metrics, and managing terminology version updates. An advisory committee — with representation from the business functions that depend on the ontology — provides guidance on priority, scope, and fitness for purpose. A change control process — with defined request, review, approval, and publication stages — ensures that changes to the ontology are documented, reviewed for downstream impact, and communicated to dependent systems before deployment.
Quality Metrics for Knowledge Assets
Quality monitoring for ontology-managed knowledge assets requires metrics that are specific to the knowledge representation context. Coverage metrics measure what fraction of the relevant domain concepts are represented in the ontology, and what fraction of the data being annotated maps successfully to ontology identifiers. Accuracy metrics, measured against a curated gold standard, track annotation precision and recall over time. Currency metrics track the lag between source data changes and knowledge graph updates. Consistency metrics identify logical inconsistencies detected by automated ontology reasoners. Tracking these metrics over time, and acting on deteriorations before they affect downstream applications, is the core of an effective knowledge governance programme.
Integration with Data Quality Frameworks
Ontology governance should be integrated with the broader data governance and data quality frameworks that pharmaceutical organisations operate under ICH E6 (Good Clinical Practice), ICH Q10 (Pharmaceutical Quality System), and applicable data integrity regulations. The ontology and knowledge graph are data assets that require the same level of documentation, access control, change management, and audit trail as any other regulated data asset. Framing ontology governance within the existing regulatory data quality framework — rather than as a separate technical activity — makes it easier to resource, justify, and sustain over the multi-year timescale required for a knowledge asset to deliver its full value.
