Entity research
Early access

Know the subject before writing about it.

A keyword is a string of words. An entity is a thing with names, parts and relationships. The engine researches the entity behind each page, then the trade's own guidance.

The knowledge pack

What each page gets

Sources in a page's knowledge pack
SourceWhat it contributes
WikidataThe page's main entity and the things it works on, with their names and relationships
WikipediaUp to fifteen sections on the subject, stored with the revision they came from
ESCOUp to six skills from the European skills framework for the trade's occupation
Trade guidancePassages from governing bodies and public guidance, word for word
Right entity

Catching the wrong match

Search for "roof" and a database returns people, films and places. The engine proposes entities with warnings. Matches that are not concepts, such as people and works of fiction, are dropped. A person approves the final choice.

  • Plural and singular names are treated as one
  • Trade words in the business name don't skew the match
  • Each choice is recorded with who approved it and when
Trade guidance

Public sources, one trade at a time

Each trade has its own list of sources: government, regulators and public bodies first. For solar that includes GOV.UK, Ofgem, HSE and the Planning Portal. For dentistry it includes the NHS.

Each page is captured once, split into passages under their headings, and fingerprinted. A page gets the passages that name its subject, so it can cite them.

Pages name only governing bodies in the text. Charities, trade associations and companies are cited in the evidence trail only. Competitors are never named.

Research the subject, not the keyword.

See what the auditor finds missing from your page.