I've found the author of Voynich manuscript. He was Giovanni Aurispa.

[tavie]

Yes i understand your worry. But that is the reason keywords are very important. Sometimes they are easy to find, I will give one example : From the picture below one of the keywords is the first word Balaustiis ( pomegranate flower). So the plant is pomegranate. 

But your system does not produce "balaustiis".  The EVA is kchsy, and your system produces blsts.  It's then your brain that turns it into a recognizable word by adding in vowels at your discretion.  You have free rein to add whatever vowel you like, including dipthongs, at any part of the word, including at the beginning and the end until it produces a word you are happy with.  It looks like you can also double consonants if you choose.  

This means you could have chosen 

• "balistas" (and a whole host of different inflections), 
  
• "abluistis";
  
• "bulliistis", and
  
•  "ebulliistis". 
  

You've gone for "balaustiis" because a pomegranate flower makes more sense in a herbal manuscript than the others and you want your solution to be correct.  But this is confirmation bias.  Someone who is using your system but is not personally invested in it being true would not necessarily settle for "balaustiis" being the word.

The problem is worse for words like chor and chol as Marco showed.  The vowel wild cards - and indeed the ability to insert any vowel into any place of the word, including the start and end of the word - show that there are far many more possibilities than the selection Marco chose.  This gives your confirmation bias free rein to pick the word you think is most suitable.  Your system is not producing the translations:  it is your brain that is doing the heavy lifting by going through all the possibilities and picking one.  Different users of the system would pick different words. 

You can see the same issue in any of the rival solution threads I linked to.  The Turkish thread in particular focuses on "keywords" that they argue would be too much of a coincidence to have randomly popped up, given the expected content of the folio in question.  In fact, they say it is "mathematically impossible".  But all things are possible when you have so much freedom in the system. 

The biggest problem for your system, however, is that even with all these degrees of freedom, the output is nowhere near being viable Latin. 

My advice to you is similar to before:  have a read of at least some of the discussions on rival solutions; see what reasons people bring up in their criticisms; and consider how those apply to your own system and methodology, and what you can do differently.  My biggest piece of advice would be not to focus on solving the Voynich but instead on adding to the information we know about how it works.