Trace the material evidence
An explanation should identify which observations materially support the dimension on screen: subject, predicate, source, date and a retrievable reference. A list of links without a connection to the attribution is not enough. Neither is a plausible story generated after the fact. A person should be able to inspect the material that made a capability or relationship worth attention.
Preserve how the interpretation was reached
A derived conclusion depends on questions, state contracts and versioned policies. An explanation should make clear which dimension was evaluated, which information was considered and what was deterministic or inferred. Structured model output is neither an explanation nor an authority. If the system cannot reconstruct the path, it should not present the attribution as an intelligible role.
Expose relevant limits and contradictions
A balanced explanation retains evidence that challenges the reading, stale sources, partial coverage and UNKNOWN values. Selecting only supportive items can turn an uncertain derivation into an apparently certain story. Uncertainty should reflect sources and conditions, not an opaque model confidence or the commercial outcome someone hopes for.
Keep explanation separate from canonical admission
A POTENTIAL association may have enough basis to appear as an investigable question even though it is not a FAXT or admitted relationship. The user retains authority to accept, reject or reinterpret its business meaning. The admission boundary remains separate: an explainable basis supports inspection but does not automatically authorize a write to AXIGLAND truth.
Conceptual example
Hypothetical: a Signal proposes investigating whether an observed capability may fit a public project. Its explanation opens the capability and project sources, shows their dates and states that eligibility and a relationship are unproven. The person can see why the Signal appeared and what remains open.
Scope and limits
An explanation cannot repair an unsuitable source or prove a derived judgment correct. It must preserve attribution, time and material contradictions. If AXIGNAL cannot reconstruct an adequate basis for a visible attribution, it should remain UNKNOWN/UNRESOLVED or prompt investigation rather than receive a post-hoc narrative.
Editorial basis
This page explains product doctrine and boundaries. It does not demonstrate source coverage or observed outcomes.
Read the product modelExplore knowledge boundaries
Explore AXIGNAL