How CineQuiz produces its content
The whole path, from the data to the published text.
Where the data comes from
The catalogue comes from TMDb: title, synopsis, cast, direction, genre, runtime, budget, box office, rating and vote count. Awards, nominations and source work come from Wikidata. None of those numbers is generated — they all come from the database, which is why they can be checked.
How a text gets written
Every column starts from a record built out of the catalogue. A columnist — a persona with its own thesis about cinema — receives that record and writes. The text then goes through automatic checking, and is only published if nothing is flagged. Whatever fails goes back for human review.
What the checking looks for
A scene described by someone who has not seen it. Quoted dialogue. The ending revealed, even when the synopsis reveals it. A character the synopsis does not introduce. A claim about reception with no figure behind it. A number, a year or a ranking position the record does not hold. Clichés and generic essay constructions. Contradiction with the record itself. Vocabulary from the wrong language.
What the columnists never do
They have not seen the film, so they do not describe a scene, a shot, the score, the cinematography or the editing. They do not quote dialogue. They do not reveal the ending. They state nothing that is not on the record — when a fact is missing, the text says so.
Disclosure
Every text signed by a columnist carries the AI badge next to the byline, at the top of the page, linked to that columnist's page. No columnist text is published without it.
