Word based data compression schemes

Documents, papers, and reports contain large amounts of redundancy. This redundancy can be minimized by data-compression techniques to save storage space or to increase transmission efficiency. Several data-compression algorithms that are character based have been proposed in the literature. In English test files, however, the natural units of repetition are words or phrases, rather than characters. Three different source models for word-based data compression are proposed: move to front, frequency to front, and alpha-numeric to front. Their principles and methods for encoding their gathered data context are presented. Results of compression ratios obtained are included and compared. Comparisons with the performances of the Lempel-Ziv algorithm and fourth-order arithmetic encoding are also made. Some ideas for further improving the performance already obtained are proposed.<
>