AI Analytics Glossary

In analytics, an AI hallucination is a confident, plausible answer that is wrong, such as a metric calculated the wrong way or a number the data does not support.

Analytical drift is when the same business question produces different answers depending on who asks it, how it is phrased or which tool answers it.

Business context is the knowledge that gives data its meaning in a specific company: how metrics are defined, which rules and exceptions apply, what terms mean in each team, and how the business actually operates.

A business data product is a curated, governed set of data built for a specific business purpose, with a clear owner, documented meaning and quality standards, so people and AI can use it with confidence.

Context engineering is the practice of designing and supplying the information an AI system needs at the moment it answers, such as definitions, data, rules and instructions, so its output is accurate and grounded.

A context layer is the business knowledge that AI systems draw on to interpret data correctly: definitions, rules, relationships and terminology, kept in one place so every AI tool uses the same understanding of the business.

Analytics that understands the business context behind the data and delivers trusted answers where people work.

A contextual semantic layer is a semantic layer that also captures business context, such as why metrics exist, which rules and exceptions apply, and how terms differ across teams, so both people and AI interpret data the same way.

Conversational analytics lets people ask questions about data in plain language, by chat or voice, and receive answers, charts and explanations without writing queries or building dashboards.

A knowledge graph is a network of entities, such as customers, products and metrics, connected by relationships, so systems can follow connections to answer questions that span many sources.

A metrics layer, or metric store, is a central place where business metrics such as revenue, churn and conversion rate are defined once and reused by every dashboard, spreadsheet and application.

In data and analytics, an ontology is a model of the concepts in a business domain and the relationships between them, giving people and AI a shared understanding of what things mean.

Semantic governance is the practice of managing business definitions, metrics and rules as shared, owned and certified assets, so every BI tool and AI application uses the same meaning.

A business-friendly map of your data that defines metrics, entities and rules once so every tool and AI gets the same answer.

A semantic model is the structured description of a business inside a semantic layer: its entities (such as Customer and Order), how they relate, and the metrics and dimensions used to analyze them.

Text-to-SQL is the use of AI to translate a plain-language question into a SQL query that runs against a database.

A universal (or headless) semantic layer is a standalone semantic layer that sits between the data warehouse and many tools, so BI dashboards, spreadsheets and applications all share one set of business definitions.