Experimental results in Prediction by Partial Matching and Star transformation applied in lossless compression of text files

Abstract

The PPM (Prediction by Partial Matching) methods are applied before the lossless compression algorithms, as a preprocessing procedure for the text in order to use efficiently the redundancy. This paper presents 4 compression algorithms preceded by 3 different PPM methods. The compression ratio is studied in relation with the file size, its content, and the context order (the prediction degree). Some useful recommendations are given in order to achieve the best compression for any kind of text files, both in English and in Romanian, as well as in natural or artificial languages. The paper also refers to the benefits provided in the lossless compression by using a preprocessing method in order to exploit better the redundancy of the source file. The procedure is known as the Star (*) Transform and it is applied for text files. The algorithm is briefly presented before emphasizing its benefic effects on a certain set of test files chosen from basic text corpora. Experimental results were performed and some interesting conclusions were driven on their basis.