Originally designed to handle name searches better than a search engine when looking specifically for a known value. Names can be of people, places, or things - so long as the list of names is known in advance and finite. The algorithm uses n-grams of consecutive characters to find and rank the best matches to a user supplied substring.
For example, if I want to populate a drop down with the exact string `Massachusetts`, but a user types `Masachussets` which searches against a list of states and provinces from around the world - the correct one will be returned at the top of the list. Other entries are only returned if they exceed a configurable score threshold.
For example, if I want to populate a drop down with the exact string `Massachusetts`, but a user types `Masachussets` which searches against a list of states and provinces from around the world - the correct one will be returned at the top of the list. Other entries are only returned if they exceed a configurable score threshold.