Implement fast string matching algorithm that can work with Unicode alphabets described in a paper:
A Fast Generic Sequence Matching Algorithm. David R. Musser, Gor V. Nishanov
It is not really necessary to work out this specific algorithm. In some cases it is sufficient to turn a unicode string into its octets representation.
Figuring out a way to rapidly (better instantly) get an octets representation of a Lisp string would bring some benefits in general for different algorithms.
See also ticket [#8] --- Rabin-Karp implementation might benefit from fixing this issue.