Radu RĂDESCU
Transform Methods Used in Lossless Compression of Text Files

Abstract. This paper presents a study of transform methods used in lossless text compression in order to preprocess the text by exploiting the inner redundancy of the source file. The transform methods are Burrows-Wheeler Transform (BWT, also known as Block Sorting), Star Transform and Length-Index Preserving Transform (LIPT). BWT converts the original blocks of data into a format that is extremely well suited for compression. The chosen range of the block length for different text file types is presented, evaluating the compression ratio and compression time. Star Transform and LIPT applied to text emphasize their positive effects on a set of test files picked up from the classical corpora of both English and Romanian texts. Experimental results and comparisons with universal lossless compressors were performed, and set of interesting conclusions and recommendations are driven on their basis.

READ THE PDF