Would that be truly random parts or can they be identified beforehand (eg. by splitting on spaces, dashes, etc… to query on compound names or the like)?
For example do you want something along the lines of matching “Pinkman” with the criteria “ink”, or more like matching “Anna-Lucia” with the criteria “Lucia” (because Lucia is half of the compound name, bounded by “-”)?
I’m asking because I can imagine a way to do the second option: that would involve building an index that splits the firstnames and lastnames according to this limited set of rules (spaces, dashes, any relevant separator) and index each splitted element. I’m sure it can be done with views (having multiple emits), less sure how to do that with N1QL (any idea @geraldss?).
The first alternative though would indeed benefit from full text search. We are working on CBFT which aims at integrating full text search into Couchbase, but this is not yet production ready.