[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: VMs: RE: Making a vms with meaning (long)
I would no a noun, verb and adjective placement scan. Comparing to a range
of different languages. The first step is to plot the patterns in large
samples of the selected languages. Form a heirarchical tree structure of
connected word types, then let the computer do the number crunching against
the cipher text. You only need to find a few matches to start to analyse
further. Although I would get bored too quickly to even pursue this.
----- Original Message -----
From: "Gabriel Landini" <G.Landini@xxxxxxxxxx>
Sent: 28 June 2004 10:08
Subject: Re: VMs: RE: Making a vms with meaning (long)
> On Monday 28 June 2004 08:51, Elmar Vogt wrote:
> > As to solving such a code, I think you should look for words, which
> > *) appear often,
> > *) repeat often,
> > *) are placed at the beginnings and ends of sentences/paragraphs,
> > *) form characteristic sequences of two or more words.
> Yes, given that you *know* the language this was written in, one can
> some very basic structural constructs. In English this would be "and the",
> "of the", sentence initial "The" and so on. But (as in the vms) you do not
> know the language of the PT.
> > In english, a phrase like "for example" might be a good candidate for
> > last issue: While "example" overall is probably not a very common word,
> > frequently follows one particular, fairly frequent word -- "for".
> Only if that appears in the text, which you do not know. Just suppose that
> book is in English and it is about "my little pony". This would appear a
> large number of times, but how do you get to the word "pony"? Couldn't
> approach still consider correct that the 3 word pattern could be about the
> "big red balloon" or "that fat dog"?
> > Now compare those patterns/frequencies in your plaintext language
> > (whichever it is...) with the VM statistics... and do a lot of
> > et voila!
> I can send you the text and you can try it. I would say it would be
> difficult, because as in the vms you do not know the PT language.
> First you need to guess the language or propose it based on the word
> out of a large number of language samples.
> Then you can (perhaps) *guess* some of the most frequent words. I do not
> believe one can do anything else, unless you know it is a known piece of
> for which you could try matching the pattern of word sequences (brute
> > P.S.: Assuming that ciphertext words correlate to plaintext words 1:1,
> > in the light of the fact that the VM shows no punctuation -- wouldn't
> > mean that each VM paragraph is equal to one plaintext sentence? And
> > wouldn't that in turn mean... awfully long sentences?
> Punctuation was removed. But I could produced one version which maintained
> full stops.
> To unsubscribe, send mail to majordomo@xxxxxxxxxxx with a body saying:
> unsubscribe vms-list
To unsubscribe, send mail to majordomo@xxxxxxxxxxx with a body saying: