|
Current Progress: |
| Project | Est. Progress | Currently working on | Downloads |
| Main (Library/Application)
This means the application/library, for Linux and Windows, to compress and decompress using the CAT method. |
95% | Library, Console App | ZIP-File - Linux and Windows, Delphi Project, including German
Databases (English, German). (Windows: Delphi 7 [Needs QTINTF.DLL], Linux: Kylix 3) Not yet a library, console app will come soon. |
| Database Creation
An utility for the easy creation of CAT databases. |
95% | (discontinued) | / |
| Databases
Our progress in creating some basic databases (German and English) |
100% | HTTP; to download a database, you have to download all parts, i.e. meta, word, uchr, prob, folw. It is easier to get databases by selecting the Download option in the Main application. | |
| Demo Project
For demonstration purposes - compression of texts, hexadecimal display of the result, graphical display of the compression ratio. |
100% | ZIP-File - Linux and Windows, Delphi Project, including German Database (English), a few demo texts. (Windows: Delphi 6 [Needs QTINTF.DLL], Linux: Kylix 3) | |
| Detailed Description
A paper with a more detailed description of CAT in German. |
100% | OpenOffice [.sxw] | |
| Palm OS Port
An E-Book reader for Palm OS which can display CAT-compressed files on the Palm and an encoder which creates E-Books for that reader. (it will be a simpler version of CAT) |
70% | (discontinued) | / |
|
CAT is a method for compressing texts. It is the successor to
FES 1. It is superior to FES 1, because it uses
a far smaller database (ca 1:50), but achieves a better compression with it. The basic idea of FES and CAT is to compress texts by replacing words with links to a database. In FES, this was done by statically replacing words by either two or three bytes (which point to the database entry). CAT is far more flexible: the link can have from four to 20 bits, and after each word, an estimation for the words that probably follow is done (the most probable words can then be encoded with four, the least probable ones need 20 bits). |