Context-dependent stopper encoding

Abstra t. A hara ter-based en oding method is presented for naturallanguage texts and geneti data. Exa t string mat hing from the en oded text is faster than from the original text, with medium and longer patterns. A ompression ratio of about 50% is a hieved as a by-produ t. The method en odes hara ters with variable-length odewords of 2-bit base symbols. An advan ed variant is ontext-dependent, using information from the previous hara ter. The method supersedes the previous omparable methods in ompression ratio, and is omparable to the best su h methods in sear h speed.