Google Talks About How Autocomplete Predictions Are Generated
https://blog.googleAutocomplete predictions reflect searches that have been done on Google.
- To determine what predictions to show, their systems begin by looking at common and trending queries that match what someone starts to enter into the search box.
- They will also consider things like the language of the searcher or where they are searching from, because these make predictions far more relevant.
- To provide better predictions for long queries, their systems may automatically shift from predicting an entire search to portions of a search. For example, they might not see a lot of queries for “the name of the thing at the front” of some particular object. But they do see a lot of queries for “the front of a ship” or “the front of a boat” or “the front of a car.”
- They also look at freshness when displaying predictions.
- Predictions will also depend on the specific topic that someone is searching for.
They also mention things that are excluded from autocomplete predictions:
- They have systems designed to prevent potentially unhelpful and policy-violating predictions from appearing (eg: violent, sexually-explicit, hateful, disparaging or dangerous).
- The ones they miss are managed by enforcement teams that remove predictions that violate their policies.
- They will prevent showing offensive, hurtful or inappropriate queries in relation to named individuals
