Because major structured knowledge bases are often collaboratively maintained and publicly editable, creating or editing an entity record isn't like filling out a private form — it's contributing to a shared resource with its own conventions, standards, and community norms. Getting this wrong doesn't just produce a weak result; it can produce edits that get reverted or accounts that get restricted.
Mistake one: treating it like a directory listing
A common instinct is to treat an entity record like a Yellow Pages entry — name, address, phone number, a glowing description. But structured knowledge bases aren't directories, and entries that read like marketing copy or self-promotion tend to be flagged, edited down, or removed by the community that maintains these resources.
Mistake two: missing or incorrect entity type
Choosing the wrong type for an entity — or leaving it too generic — affects how every other system interprets the record. An entity miscategorised at this level can end up effectively invisible for the kinds of queries it should actually be relevant to, even if every other property is filled in correctly.
Mistake three: no relationships to other entities
An entity record with properties but no connections to other entities — no location, no category links, no broader context — is much weaker than one embedded in the graph. Relationships are often what make an entity findable in response to related queries, not just direct ones.
Mistake four: inconsistency with other sources
If the entity record says one thing and your website or other descriptive sources say something subtly different — a different founding date, a different categorisation, a different name format — that inconsistency undermines the corroboration effect that makes entity recognition valuable in the first place.