indexed similarity search

This documentation is the strongest reference in the set for substring-based matching at scale. It explains trigram similarity as a way to compare strings through overlapping character n-grams, which is valuable when exact tokenization is unreliable or when you want robust partial-match behavior for misspellings and fragments.

It is not a tutorial, and it assumes some database familiarity, but it is still a very useful conceptual reference because it ties the comparator directly to indexed similarity search. That makes it a strong citation when the question is not only how to score approximate matches, but how to do so efficiently over large text collections.