This work concerns the search for text compressors that compress better tha
n existing dictionary coders, but run faster than statistical coders. We de
scribe a new method for text compression using multiple dictionaries, one f
or each context of preceeding characters, where the contexts have varying l
engths. The context to be used is determined using an escape mechanism simi
lar to that of prediction by partial matching (PPM) methods. We describe mo
difications of three popular dictionary coders along these lines and experi
ments evaluating their effectiveness using the text fries in the Calgary co
rpus. Our results suggest that modifying LZ77, LZFG, and LZW along these li
nes yields improvements in compression of about 3%, 6%, and 15%, respective
ly. (C) 1999 Elsevier Science Inc. All rights reserved.