Semantic Layer Fundamentals
Semantic Layer Fundamentals
This domain tests whether you can read a MicroStrategy project's semantic model: how attributes, facts, metrics, hierarchies and relationships turn warehouse columns into business concepts. Expect questions on object roles, granularity, and why a metric returns unexpected totals.
Core rules
- Attributes are business entities with forms; the ID form defines the key, description forms supply labels. Two forms are not two attributes.
- Facts are numeric warehouse data at a declared grain; metrics are calculations over facts, not stored columns.
- Attribute relationships define join paths in the logical model. A missing relationship causes cross joins or inaccurate aggregation; an attribute role reuses an attribute for a different meaning, such as Order Date versus Ship Date.
- Hierarchies organize attributes for drilling; they do not enforce aggregation by themselves.
- A metric's level is determined by the attributes on the report and the metric's dimensionality settings.
Worked example
A report shows Revenue by Customer. The model has Fact Revenue at Order Line grain, Attribute Customer related to Order, and Order related to Order Line. Revenue metric = Sum(Revenue). At Customer level, the engine aggregates order-line rows up through Order to Customer. If Customer were related only to a separate Customer Address table with no path to Order Line, the result would fan out or produce a cross join. Reason: aggregation follows relationship paths, so grain must be compatible.
Traps
- Treating an attribute form as a separate attribute.
- Assuming a metric stores data; it is a definition.
- Forgetting that a report filter affects the level of aggregation.
- Confusing a hierarchy with a relationship: drilling uses the former, joins use the latter.