Graph Equivalence

Semantic reconstruction can preserve meaning even when the serialized output is not byte-for-byte identical to the original representation.

For that reason SKC distinguishes structural equivalence from raw file identity.

Example

Representation A:

System ──produces──> Evidence

Representation B:

Evidence <──produced-by── System

Whether these representations are equivalent depends on the declared semantic contract.

SKC must not silently assume equivalence.

Graph-level validation

A graph-equivalence check can consider:

• node identity
• node type
• edge identity
• relation direction
• relation semantics
• required attributes
• cardinality
• canonical ordering rules
• allowed normalization

Why byte identity is insufficient

Two valid semantic states may have different:

• JSON ordering
• whitespace
• serialization layout
• internal identifiers

while preserving the same declared graph structure.

Conversely, two files may look textually similar while encoding materially different semantic relations.

Contract requirement

Graph equivalence must therefore be evaluated against explicit rules.

Graph A
   ↓

Equivalence Contract

   ↓
Graph B
   ↓

PASS / FAIL

Result

GraphEquivalence is one of the important SKC validation metrics.

It should be interpreted together with reconstruction fidelity and determinism.

Next

Continue with Metrics →.