It has become obvious and generally accepted now that AI agents can’t function properly without enough context. As organizations scale up their AI use and strive for new, agentic workflows, various technical and architectural solutions for context management have emerged. But what do these semantic layers, context planes, and knowledge management systems actually contain? And where do we get all that context from?
In this talk, we will focus on what information is actually needed, instead of how that information is stored or processed. Zooming out, we’ll find out that in the end all the “context” or “knowledge” is made of very simple basic elements – things, definitions, and relationships – that are very familiar for those of us coming from a data modeling background. We will look into the daunting world of Knowledge Graphs and Ontologies through this very practical lens, which allows us to avoid getting tangled in standards and syntaxes and lets us concentrate on the important part: the knowledge itself.
- What is this “context” everyone keeps talking about
- Things, definitions, and relationships – back to basics
- Metamodels – what do we need to know about
- Using Conceptual Modeling as a context discovery method
- Recap on Knowledge Graph and how it relates to ConceptualModels
- What goes where in the Knowledge Graph pyramid: glossaries, ontologies, and instances
- Metadata graphs vs. “actual” graphs – differences in scale and use cases
- Managing and maintaining knowledge in the modern Enterprise.
