Consensus building means helping people to reach agreement on an issue that affects them all, by following a collaborative decision-making process with clear ground-rules. In the world of data, these methods are used to build consensus around the design of ontologies as preparation for the roll-out of AI services.
What is an ontology?
An ontology is a structured framework that sets out the rules and relationships that govern a domain’s knowledge, making it machine-readable for AI services. Ontology standards enable data to be united and interpreted consistently across systems, known as semantic interoperability, where the precise meanings of key words (controlled vocabulary) are agreed on.
Overall, ontologies enable an AI service to improve its accuracy and trustworthiness by ensuring that it only uses verified data, while demonstrating strong contextual understanding of the domain’s rules and standards.
Why does vertical AI need consensus?
With access to structured data, vertical AI – a highly optimised, industry-specific model – can perform critical and high value tasks with fewer errors or hallucinations. Ontologies are therefore key to faster and more reliable domain intelligent systems, which are in growing demand.
Before benefiting from advances such as vertical AI agents, however, stakeholders need to sign off on standardised definitions of those ontological rules and concepts. They need to achieve consensus.
How do you build consensus for an ontology?
Traditionally, the process of building industry-wide consensus around an ontology tends to involve face-to-face conferences and roundtables. Progress can be slow however due to conflicting priorities, conceptual disagreements, and the complexity of aligning multiple stakeholder interests.
A mediator, or change agent, is therefore appointed to supervise the process of consulting stakeholders and prioritising the key issues to address. Participants go through rounds of discussion to arrive – hopefully – at a solution that can be supported by everyone.
The Delphi Method is one of the best known approaches to consensus building. A 2024 study set out how the method helped achieve consensus among global experts for standardising phenotypic annotations in rare disease diagnosis, while updating the Human Phenotype Ontology.
One way to get started on agreeing an ontology is to ask your domain experts three key domain questions.
What is Ontology as a Service (OaaS)?
Ontology as a service (OaaS) is designed to make ontologies more accessible by offering an easier, faster and more flexible digital environment for developing domain intelligent systems. Ontology software platforms often provide APIs, visual modelling tools, and cloud-based environments for creating, editing, and sharing ontologies. This supposedly enables non-technical experts and smaller businesses to get more value out of their knowledge using vertical AI.
How does OaaS improve consensus building?
OaaS provides collaborative tools to accelerate consensus building. It enables anonymised feedback to avoid personal conflicts or ‘groupthink’. It introduces deadlines and nudges to keep momentum going. Features like version control, audit trails (XAI), and rollback options make it easier to iterate and adapt ontologies as consensus evolves. Overall, OaaS aims to remove complexity from the user interface, improve engagement, and raise the chances of getting a working ontology over the line.