There has been some work done on text-filters like the one you mentioned. One of the tricks I can recall is that they ‘grouped’ frequently used 2-3 character combinations and replaced them with a single ‘unused’ character. (eg, ‘the’ can be replaced by 0×01. the is used very frequently in the, them, their, there etc and 0×01 is never used in text files.) Another one was about optimizing encoding of new-line characters and special handling of capitalization of first characters after a ‘.’ period.

You might want to look up these papers. Let me know if you can’t find them and I will try to see if I have them in my old archives.

On the other hand …

I’ve been playing with something distantly related to this — recognizing “compression” in English text (acronyms, contractions, standard patterns for making words plural, other patterns for adding “ing” or “ed” suffixes, etc.).

So my decompressor has 2 stages:

First decompress to a “verbose” text analogous to the “.pcf file”, which has text something like “cooky +s”.

Then “compress” using standard English patterns:
“The plural of *y is *ies”.

The “bit-for-bit identical” is important for this application, because I want the result — after the second stage — to be bit-for-bit identical with the original English text.

thanks

Best way to protect your rights would be to get a patent on your method, but it will have to be something really new. Best of luck.