On the regex suggestion, once upon a time, I did a project which used Regexp::Assemble to take a long list of rexeges and combine them all into a single, monstrously unreadable, regex that would tell me in one shot whether any matched. I seem to recall hearing at some point that similar functionality had been added to the Perl core, but don't recall details.

In any case, if you're looking for fixed strings rather than regex patterns, substring matches, etc., I would still expect hashes to be faster, but something based on Regexp::Assemble or a similar technique could be close enough to warrant a benchmark. And you might still want to go that route even if it's a little slower in order to gain the extra flexibility in how you match.