Align entities and attributes before combining them
A trade name may refer to a brand, subsidiary, or different legal entity. Before combining data, check identifiers, addresses, aliases, and dates, and document ambiguity. Then ask what each attribute means: a shared address might be a co-working office, a registered address, or stale information. Keep records separate when there is not enough basis to resolve them.
Assess each source's authority and purpose
The most useful source depends on the information you are checking. A corporate website may express declared capabilities; a registry may document a filing; a certifier may confirm a certification within its scope. The same source may suit one predicate and be weak for another. Record when it was observed and which text, field, or reference supports the extraction.
Do not turn overlap into a relationship edge
Two organizations operating in the same sector, region, or possible supply chain does not show that one sells to the other. An event participant list does not establish a commercial partnership. An observed relationship needs specific, sufficient evidence; compatibility might, with context, support a potential hypothesis that must be labelled as such rather than presented as an existing tie.
Show conflicts and gaps instead of forcing a synthesis
If sources use different names or incompatible dates, keep the conflict visible and state what is needed to resolve it. Do not choose the data that makes the story look most coherent. A cross-check should show what came from each reference, what was combined, what stayed separate, and what cannot be answered. Another person can then review the connection without accepting an opaque conclusion.
Conceptual example
Hypothetical example: a public directory says a workshop makes components, and a procurement record shows an entity bought a category of parts. If neither the awarded manufacturer nor a relationship between those subjects is identified, the overlap only supports asking whether a connection exists. It does not justify describing the workshop as that entity's supplier.
Scope and limits
Public data can be incomplete, duplicated, outdated, or subject to different reuse rights. Publication alone does not grant permission for automated collection or guarantee accuracy. Aggregated attributes do not replace evidence of a specific relationship. If identity or the connection remains unresolved, express UNKNOWN or abstain from asserting it.
Editorial basis
This page explains product doctrine and boundaries. It does not demonstrate source coverage or observed outcomes.
Read the product modelExplore Knowledge
Explore AXIGNAL