Fast text search queries can be made even faster with indexing on lexemes inside compound records of tsvector format. RUM index is an open-source PostgreSQL extension. It represents a big improvement of GIN index and it can index lexemes with additional information e.g. tsvector lexeme weight-mark. So it can support tsvector capabilities more.

Until recently it was needed to recheck results of weight-containing queries by table. My modification (2020) is to make the processing of this kind of queries index-only and therefore much faster.

Also, I will describe and provide benchmarks for different usage cases for fast text search. We'll see how RUM index can improve the performance and compare it with PostgreSQL internal GIN index.