The Voynich Ninja

Made an edited version of ZL transcription. "ZL_full_clean_01.txt"

226328  chars (excluding newlines)
38696  words
5389 lines

constructed like this:

ivtt.exe -x7 ZL_ivtff_1r.txt ZL_full_clean_01.txt    -cmdline to create file

uncertain spaces    = spaces    -done by IVTT
uncertain readings  = 1st one   -done by IVTT

rare chars i.e. @123;  = '?'    - 166 replaced
apostrophes        = ''           - 97 replaced

???                ='?'            -31 replaced
??                ='?'            -09 replaced
?  333 occurrences

words containing or consisting of '?' removed

Has max text file size changed ? Oh well here is a pastebin. ( 231.55 KB )
You are not allowed to view links. Register or Login to view.

(17-09-2020, 05:33 PM)Koen G Wrote: You are not allowed to view links. Register or Login to view.No but seriously, if one could train one of those "AI" to optimize Voynichese that would be interesting. Do you think this kind of software is accessible nowadays? Not to me obviously but perhaps to experienced programmers.

Yes, stuff like googles tensorflow is freely available and most of the recent papers have supplementary githubs which include all their code.
However it's not so easy to put it all together, way beyond my sikill level.

(17-09-2020, 05:33 PM)Koen G Wrote: You are not allowed to view links. Register or Login to view.Do you think this kind of software is accessible nowadays?

There was this thread You are not allowed to view links. Register or Login to view.

It's a pity that Hauer and Kondrak's work has been damaged so badly by their naive usage of Google Translate. But what they tried to do was largely sensible, when one begins to think of complex systems like anagrams or verbose ciphers. I doubt that these are scenarios where a solution can be found manually, considering that the target language is also unknown.