nablator > 10-06-2019, 03:36 PM
(10-06-2019, 02:59 PM)ReneZ Wrote: You are not allowed to view links. Register or Login to view.This results in a situation where similar words appear near each other.This works well for the first time they appear, but the next time they will be less likely to be near each other, so I suspect that it will be hard to maintain this good result on a much longer text. Spikes on first correlated appearances will not offset the overall uncorrelated appearances. It would be interesting to check whether uniquely appearing words (hapax legomena) are more or less line-distance-edit-distance-correlated than other words, also in the VMS.
ReneZ > 11-06-2019, 09:44 AM
ReneZ > 11-06-2019, 10:30 AM
nablator > 11-06-2019, 10:47 AM
nablator > 11-06-2019, 11:18 AM
(11-06-2019, 10:30 AM)ReneZ Wrote: You are not allowed to view links. Register or Login to view.So here is one more attempt.pliny_norm:
I used the first approx 10,000 words of Pliny's natural history.
It has more than 3999 word types, so I added the Roman numeral Q to represent 5000.
I also introduced an alternative representation of the numbers (not Voynich-like).
Here are the files:
You are not allowed to view links. Register or Login to view.
You are not allowed to view links. Register or Login to view.
You are not allowed to view links. Register or Login to view.
Second edit: problem should have been solved.
Koen G > 11-06-2019, 12:35 PM
nablator > 11-06-2019, 12:48 PM
(11-06-2019, 12:35 PM)Koen G Wrote: You are not allowed to view links. Register or Login to view.Noob question: if your words are all very short, won't this increase the likelihood of lower values?It will, but values don't matter, only the relative variation between near values and far values matters. A list of pseudo-Voynichese words generated from earlier words by modifying and adding glyphs within constraints should work as well or better than Roman numerals.
ReneZ > 11-06-2019, 12:56 PM
Koen G > 11-06-2019, 01:48 PM
Koen G > 11-06-2019, 01:55 PM