Data preprocessing by sequential pattern mining for LZW

LZW is a lossless data compression algorithm which has been incorporated as the standard of the Consultative Committee on International Telegraphy and Telephony. In addition, LZW is used to create GIF, TIFF and PDF files. In this paper, we propose an improvement to LZW using ideas from sequential pattern mining. The goal of this area is to find all the maximal frequent sequences (MFSs) which are sequences that appear at least β times and they are not subsequences of any other MFS. We preprocess the data using an algorithm for searching all the MFSs to manage the MFSs as part of the dictionary of LZW, according to the frequency of the MFS. This modification allows us to propose a new variant of LZW algorithm. Some experiments with text files, showing the compression rates of the proposed algorithm, were performed.