An Evaluation Tag records how a Fact captured by a Fact Tag was evaluated under the Applicable Governance Basis represented by a Policy Tag. An Evaluation Tag is therefore neither Fact nor Policy. It is Evidence of the Evaluative State formed by relating the two.

Evaluation requires an Evaluation Configuration that defines how the assessment is performed. This may include Applicable Evaluation Criteria, Evaluation Dimensions, Threshold / Sensitivity, Weight / Criticality, and Aggregation Rules. Because the same Fact may produce different Evaluation Results under different Criteria or Configurations, the system must preserve not only the result, but also the Configuration through which that result was formed.

Criteria define what is evaluated, while Dimensions represent the Evaluation Factors that constitute each Criterion. Threshold / Sensitivity determines which states or changes are treated as Signals, while Weight / Criticality defines the relative importance and Governance Significance of individual Criteria or Dimensions.

Organizations may adjust the Configuration according to Risk, Domain, Operational Context, and the AI System concerned. However, this discretion remains constrained by Applicable Policy. Evaluation Configuration cannot be used to circumvent or invalidate Applicable Policy.

Evaluation Results may include Criterion-level States, Dimension-level Results where necessary, an Overall Evaluation State, and Exceptions or Unresolved Conditions. Where multiple Criteria apply, each is evaluated independently and then integrated into an Overall State according to Weight, Criticality, Policy Alignment, and Aggregation Rules. The Overall State is therefore not determined through simple majority voting or mechanical averaging, and the applicable Aggregation Rules themselves must remain verifiable.

Evaluation also does not need to artificially eliminate Uncertainty. Where required Facts are missing, Criteria conflict, or uncertainty exists near a Threshold, the condition may be preserved as Grey, Exception, or Unresolved Condition. This allows Evidence to capture not only what was evaluated, but also what could not be determined.

An Evaluation Tag therefore records more than a final State. By preserving the Evaluation Path:

Criterion → Dimensions → Threshold / Sensitivity → Weight / Criticality → Criterion-level State → Overall State

it becomes possible to reconstruct why a particular Evaluative State was reached. While the Fact Tag provides Evidence of What Is and the Policy Tag provides Evidence of What Governs, the Evaluation Tag provides Evidence of How It Is Evaluated, making the Evaluation Process connecting Governance to the subsequent Decision verifiable.