Everyone is talking about context when it comes to AI
Ash Srivastava - June 25, 2026
Everyone is talking about context when it comes to AI.
Have you ever wondered what all these terms mean?
Data management terminology can easily blend together.
Let's look at what the most commonly talked about terms:
Metadata: The underlying substrate. It is the background data about an asset—like a serial number, creation timestamp, or lineage path—that makes information findable, auditable, and governable.
Semantics: The meaning attached to data. It translates raw values and data types into recognizable, real-world concepts so that both human analysts and LLMs understand exactly what a term like "customer" or "revenue" means.
Taxonomy: A hierarchical, parent-child classification. It structures concepts into clean, top-down trees (e.g., Electronics ➡️ Computers ➡️ Laptops), narrowing down scope so organizations can ask clear, aggregate questions.
Ontology: A formal model of complex, multi-directional relationships. While taxonomy handles strict hierarchies, ontology maps out how completely different entities, attributes, and concepts connect across an ecosystem (e.g., mapping how a user, an asset, a location, and a policy interact).
Knowledge Graph: The ontology populated with real-world data assets. It anchors structural frameworks to live reality, enabling advanced AI agents to perform complex, multi-hop reasoning across interconnected data nodes.
Context: Real-time situational awareness. Sitting at the pinnacle of the architecture, it uses continuous data streaming infrastructure to synchronize your models with the immediate state of the operational world.
💾 Save this
♻️ Repost so a leader in your network sees this framework.
🔔 Follow Ash Srivastava for daily insights on Data and AI Governance.
Credit:
First shared on LinkedIn.