[Article] Demystifying the Voynich manuscript using computational linguistic techniques

[Torsten]

There is a new paper published about the VMS: You are not allowed to view links. Register or Login to view.

The author is Kevin Farrugia from the University of Malta.

The paper describes an experiment that takes the proposed classification by Dr Lisa Fagin Davis as ground truth and puts it to the test. The text is split into an equal number of pages per scribe; taking into consideration three of the five scribes due to scribes 4 and 5 having much less data.

Farrugia describes his results as:

In both experiments, with trigrams and bigrams, all classifiers have very high F1 scores for scribe 1. ... This information is reflected in the confusion matrices in which, with the help of the heatmap, it is easy to see that pages labelled as scribe 1 in the ground truth are rarely classified as the other two scribes. The tendency among the classifiers is to classify pages labelled as scribe 2 to be scribe 3, and vice-versa.

Not surprisingly the paper lists "the trigram 'edy' and the bigram 'ed'" as criteria to distinguish between scribe 1 and scribe 2/3 (or between Currier A and Currier B).