Embodiments of the invention include a dictionary based data compression method, apparatus and system that is not based on either the LZ77 compression algorithm or the LZ78 compression algorithm, but includes many features of the LZW compression algorithm. The data compression method includes creating a mapping table of the messages in the alphabet of messages to a corresponding plurality of codewords, maintaining a dictionary including a mapping table of a first codeword and a second codeword to a new codeword, reading an input ensemble including a plurality of messages, converting the messages to an input codeword using the mapping table, and outputting the converted codewords as an output ensemble of compressed data. Unlike conventional data compression methods, the dictionary is generated from the output ensemble only, and is not based on any input messages. Therefore, the dictionary more quickly builds to define longer sequences of messages compared to conventional data compression methods.