What’s been done up to now:
The most common use-case for taxonomies and ontologies is to improve precision and recall in information retrieval. This is typically achieved by curating knowledge organization systems that contain single-source-of-truth enterprise-controlled vocabularies. These vocabularies provide consistent metadata values for classifying and searching content across disparate content management systems. The structure of taxonomies and ontologies are also often used in information architecture to support knowledge discovery via navigation menus, faceted search, and content recommendations.
What’s now possible:
By combining taxonomies and ontologies with AI, novel use-cases emerge. Taxonomies and ontologies can enumerate decision points and criteria, and be linked together to define process triggers, dependencies and consequences. AI agents then reference the taxonomies and ontologies to provide definitive instructions on how to run business processes.
Agentic ontologies are therefore graphs structured with process-oriented semantic concepts, relationships and properties. They provide the framework for how AI agents perceive, reason, make decisions, and act. Agentic ontologies provide AI agents with the enterprise domain knowledge to manage complex interactions with humans and other systems. The transparency of the ontology graph empowers human agency over the outcomes of AI agents.
Changing what we do:
While taxonomies help support AI agents, AI agents also support taxonomies.
Generative AI and Agentic AI are rapidly changing the way that many of us work. If one is engaged in developing taxonomies and ontologies, or managing metadata and classification, then embracing AI will be crucial to the success of your team.
AI tools can deliver major productivity and quality gains. This talk will explore how Generative AI supports advanced information extraction techniques to expand and enrich the structured knowledge in taxonomies and ontologies.