AI First Data

Glossary

Metric definition

The agreed business meaning, formula, filters, time logic, and owner for a measure.

Source: ProductFeatureSet

Semantic layer

Governed business meaning placed over raw data so analyses use consistent definitions.

Source: ProductFeatureSet

Query lineage

The trace from an analytical result back through query logic, transformations, and source data.

Source: ProductFeatureSet

Read-only query

An analysis operation that retrieves data without changing the underlying source.

Source: ProductFeatureSet

Schema profile

A structured summary of fields, types, relationships, completeness, and other data characteristics.

Source: ProductFeatureSet

Anomaly

A result that differs from an established expectation and requires contextual investigation.

Source: ProductFeatureSet

Assumption

A stated condition used in analysis that a reviewer can inspect and correct.

Source: ProductFeatureSet

Data freshness

How recently a source was updated relative to the business question being answered.

Source: ProductFeatureSet

Reconciliation

The process of comparing definitions or results and resolving meaningful differences.

Source: ProductFeatureSet

Row-level permission

An access rule limiting which records a person or workflow may use.

Source: ProductFeatureSet

Sensitive field

A data element requiring restricted handling, masking, or exclusion under policy.

Source: ProductFeatureSet

Validation check

A test of totals, joins, filters, definitions, or outliers before an analysis is accepted.

Source: ProductFeatureSet

Reporting period

The explicit time window covered by an analysis or report.

Source: ProductFeatureSet

Approval queue

A visible place for a named reviewer to edit, reject, pause, or approve proposed work.

Source: ProductFeatureSet

Audit trail

A reproducible history of inputs, assumptions, queries, reviews, and outcomes.

Source: ProductFeatureSet