Text compression using lexicographic permutation of binary strings

Abstract

Text messages are generally encoded by performing table look-up on fixed length code tables. In this paper, a lossless text compression algorithm which works on the principle of entropy reduction is proposed. Characters in a text message in any language are generally encoded using a binary string with a Unique Lexicographical Rank (ULR). A corresponding Maximum Rank(MR) for any binary string can be computed using lexicographical permutation. Reducing the MR of the binary string results in considerable reduction in the number of bits to be transmitted. MR reduction is achieved in this work by using character frequency based encoding models. Uni-gram, bi-gram and tri-gram models are used herein. Experiments on text compression are conducted on the Calgary Corpus and Project Gutenberg databases. Experiments on text compression are conducted on the Calgary and The Project Gutenberg corpus. Results indicate a significant increase in compression ratio.