Glossary
Key terms used across the Analytical Readiness Framework and Knowledge Base.
Semantic drift
A gradual change in what a metric or field means, often caused by hidden logic changes or shifting business rules.
Canonical metric
The single, authoritative definition of a business metric that all reports and AI answers should reference.
Measure singularity
The principle of keeping one measure per business concept to prevent conflicting answers.
Filter context
The set of filters, slicers, and relationships that determine which data is included in a calculation.
Context volatility
When small changes in filters or relationships cause large, unpredictable changes in results.
Lineage
A traceable path from a reported number back to its source tables, columns, and transformations.
Attribution
Assigning impact or credit to factors that contributed to an outcome.
Contribution analysis
Breaking a total metric into the components or segments that drive it.
Dimensionality
The number and structure of dimensions used to describe a fact, such as product, region, or time.
Grain (level of detail)
The smallest unit at which a fact table records events, such as order line, invoice, or day.
Star schema
A model design with a central fact table connected to dimension tables.
Snowflake schema
A model design where dimensions are normalized into multiple related tables.
Role‑playing dimension
A dimension used for multiple roles (e.g., Order Date vs Ship Date) in the same model.
Surrogate key
A synthetic key used to uniquely identify a record when natural keys are unstable.
Inactive relationship
A relationship in the model that is defined but not used unless explicitly activated in a measure.
Bidirectional filtering
A relationship setting where filters flow in both directions, often increasing ambiguity.
Ambiguity
When multiple valid paths or definitions exist for a metric or filter, leading to inconsistent answers.
Many‑to‑many relationship
A relationship where multiple records on both sides can match, often requiring a bridge table.
Calculation group
A Power BI feature that applies a set of calculations across multiple measures.
Measure branching
A pattern where complex measures are built from reusable base measures.
Time intelligence
Logic for comparing time periods (e.g., last month, year‑over‑year).
Default aggregation
The automatic aggregation applied to a field when no explicit measure is used (e.g., SUM).
Metadata density
The coverage of descriptions, annotations, and definitions across model objects.
Semantic contract
A documented agreement describing how a metric should be interpreted, queried, and validated.
Retrieval
Fetching the right context, metadata, and constraints to inform an AI answer.
Grounding
Anchoring AI responses to verifiable data and model definitions.
Deterministic query
A query that yields the same result every time given the same data state.
Explainability
The ability to justify a result by showing drivers, sources, and assumptions.
Drivers
The factors that explain changes or differences in a metric.
Variance decomposition
Breaking a change in a metric into contributing components (price, volume, mix).
Cohort
A group of entities that share a common starting event or time period.
Segmentation
Dividing data into meaningful groups for analysis and explanation.
Sensitivity
How much a metric changes when a specific input changes.
Outliers
Data points that are unusually high or low compared to typical values.
Null semantics
The meaning of missing values (e.g., unknown vs not applicable).
Unit semantics
The units associated with a value (e.g., dollars, percent, count).
Currency normalization
Converting values into a common currency or exchange rate basis.
Hierarchy
An ordered structure such as year → quarter → month or country → region → city.
Slicer
A UI control in Power BI that filters data by a dimension.
Row‑level security (RLS)
Rules that restrict which rows a user can see based on identity or role.
Measure inventory
A catalog of all measures in the model with definitions and usage.
Metric sprawl
Proliferation of similar metrics that creates confusion and conflicting answers.
Relationship path
The chain of table relationships that connects a dimension to a fact table.
Bridge table
A table used to resolve many‑to‑many relationships by providing a single path.
Semantic annotation
Structured metadata that encodes meaning such as grain, decision, or owner.
KPI
Key performance indicator; a measure tied to a business goal.
Explanation template
A standardized structure for explaining why a metric changed.
Context test harness
A set of repeatable queries that verify context and filter behavior.
Semantic compatibility
How well a model’s definitions align across tools and AI systems.
Retrieval context
The specific metadata, filters, and constraints passed to an AI system for answering.