How does it work? Specify as many fields as you like for text indexing Each field can have an associated weight Individual result documents are relevance-ordered by score Supports basic boolean queries: OR, AND, NOT, PHRASE Operational considerations Works with sharding and replication You can have one text index per collection The text index has an associated default language Under the hood Optional search arguments include project and filter It’s a command: results are wrapped in a single return document Text strings generate stemmed, non-stop-word, weighted ‘terms’ The text index acts like a multi-key index on terms