Dr. Mu,
Ok - I admit defeat. This method is certainly faster than mine. There are a couple of things I would like to point out though.

You have chosen to work of words in the string rather than the keys of the hash. If someone were really doing this type of analysis on a large data set, it might be worthwhile to have multiple methods and choose the best one based on the data set. For instance - imagine that there are only 3 hash keys, but the string is 1,500 words long. It certainly doesn't make sense to use this method.

You have also assumed that the hash keys will not contain any spaces. My solution allows for this, but still has the requirement of having a word border on both ends. For insance:

my $temphash{'good boy'} = "blah";
my $string = "He was a good boy when we went to the store";
my $otherstring = "He is worthless, not a good boyfriend at all";
[download]

Your method will not work in either case. My method will correctly match in $string but will correctly fail in $otherstring.
Of course, knowing your data is what counts so that you can code for it.

My hat is off to you for such an innovative solution.

Cheers - L~R