We defined a compression algorithm called Skip Second Word (SSW) and a hierarchical partitioning schema for obtaining 256-Kbyte segments from a compressed file by 7-Zip File Manager. SSW has two main procedures. The first one is finding and skipping words, and the second one is efficiently encoding their locations (i.e., EAI) for decompression. The initial step is to find two identical 16-bit words within a 64-byte page. If there is such a page having an encoding advantage, then we can save the first word in the output file and compress the other one by skipping. We developed Recursive, Incremental, and Iterative programs based on the SSW algorithm. We also implemented two non-SSW programs, Heuristic and Sequence, obtaining sub-files for further recompressions using 7-Zip. We tested all programs and chose the best versions of them to build a compression suite [1]. In this work, to obtain an original compression application with similar time performance, non-SSW programs are not included. Recursive and Iterative have better rate advantages than Incremental, but they are too slow. Incremental program, which iterates each segment as possible, is as fast as non-SSW ones but has no such rates. We improved it by adding an iteration parameter more and by shuffling and shifting segment contents, resulting in eight versions that are evaluated at the same run. As a result, we achieved better Space Saving rates than three SSW programs, obtaining more than 0.01% related files given in the Table.

We proposed a recompression methodology based on partitioning of a compressed data file into fixed-sized pages. We select some pages to form a separate file called the sub-file showing in the Figure. After saving the encoded addressing information related to these selected pages and the remaining pages of the input file in an output file, we then compress the newly created sub-file by using an off-the-shelf compression program as 7-Zip in ultra-option. The process ends by adding the compressed sub-file content to the output file. The recompression will be successfully done if the reduction amount obtained from the sub-file compression is higher than the size of the encoded addressing information. We partition an input file into$m$pages, therefore, there are$2^{m}$different sub-files to be formed, compressed and tested. In this work, we propose a heuristic algorithm, which is based on the selection of such a page containing identical data words, for determining pages to form a sub-file. We need only three experiments instead of$2^{m}$experiments due to three data word versions of the algorithm. The time complexity of the algorithm including a linear searching is$O(n)$, where$n$is the size of the input file in bytes. We also proposed an encoding schema to store the addressing information related to the selected pages with minimal memory cost, by designing a hierarchical partition tree of the input file in three levels [1]. Our results showed that we are able to obtain Space Saving rates more than 0.01% on chosen example compressed data files showing a few in the Table, while obtaining 3.3% for one case.