In one of the StackOverflow Podcasts (the one where guys were discussing data generation for testing DBs -- either #11 or #12), Jeff mentioned something like "reverse regular expressions", which are used exactly for that purpose: given a regex, produce a string which will eventually match said regex.
What is the correct term for this whole concept? Is this a well-known concept?
There is no such thing as a "reverse regular expression". A regular expression is just that: an expression. It expresses a language, which consists of a (possibly infinite) set of strings.
The reversal is in the use of the expression: where it is commonly used to report whether a string is in the language, it is now used to produce strings in the language.
Abstract: Recursive transition network (with the postmodernism generator as an interesting example)
One specialization would be your "reverse regex".
As to the terminology: A regular expression is a form of grammar that describes all the words belonging to a specific regular language (namely all the inputs matched by the expression).
Therefore one could call your question: "How can create a random word that matches a given regex" or "How can I obtain a random word belonging to a specified regular language".
The Perl module String::Random (in the CPAN) does this. Takes a subset of regular expressions, and does a random walk through it.
A regular expression is a shorthand notation for a generative grammar that describes a regular language.
As the name says, a generative grammar can be used to generate all words in that language. Since a “regular expression” is just a different notation for such a grammar, I think it would be correct terminology to say “generate a random string for a given regular expression”.
And as Thomas has noted, it makes therefore no sense to speak of “reverse” regular expressions.