WORD-BASED DYNAMIC ALGORITHMS FOR DATA-COMPRESSION

A new algorithm, word-based dynamic Lempel-Ziv (WDLZW) for universal (lossless) data compression, is introduced. The novel feature is that the algorithm is optimised for the compression of natural language data, in which all the spaces between words are deleted whenever copy codes or literal codes are sent out. Therefore better compression rates can be achieved. The algorithm can still compress alternative forms of data. The structure, operation and implementation of the WDLZW is described. A comparison with other algorithms when compressing a wide range of data forms is reported. For text-based information WDLZW offers attractive performance. For other forms of data, WDLZW provides compression rates similar to those of dynamic Lempel Ziv systems.