A measure of how central an identified entity is to a particular document or passage, according to a specified analysis method.
Entity salience describes prominence within content, not the overall reputation of the thing mentioned. A page may mention several organizations while mainly discussing one product. Salience analysis tries to distinguish the central subject from incidental references.
Google Cloud Natural Language v1 documents an entity salience value between zero and one for importance within the submitted text. That is a particular API output. It should not be presented as Google Search’s page score, a disclosed ranking factor or a target that every article must reach. Different tools and versions may use different outputs.
An editorial use is to investigate whether a draft stays focused. If the intended subject is hard to identify, review the introduction, examples and unnecessary digressions. Do not mechanically repeat an entity name to force a larger number; write the explanation that the reader actually needs.
Takeaway: Use salience as a diagnostic clue about focus, with the tool and input recorded. Judge clarity by reading the page as well as reviewing the output.
Example
A guide to choosing a business phone system spends most of its introduction discussing unrelated software vendors. An editor shortens that detour and explains the phone-system decision earlier. The change is justified by clarity, not a promised ranking improvement from an API score.
What to check
Compare only equivalent inputs using the same analysis method and version. A full article and a short excerpt can produce different scores, and neither result proves how a search engine ranks the page.
For more detail, see Google Cloud Natural Language v1: Entity.