Multiple-dictionary compression using partial matching

Abstract

Motivated by the desire to find text compressors that compress better than existing dictionary methods, but run faster than PPM implementations, we describe methods for text compression using multiple dictionaries, one for each context of preceding characters, where the contexts have varying lengths. The context to be used is determined using an escape mechanism similar to that of PPM methods. We describe modifications of three popular dictionary coders along these lines and experiments evaluating their efficacy using the text files in the Calgary corpus. Our results suggest that modifying LZ77 along these lines yields an improvement in compression of about 4%, that modifying LZFG yields a compression improvement of about 8%, and that modifying LZW in this manner yields an average improvement on the order of 12%.