[Article] Disambiguating Voynich Manuscript transliterations with word embeddings

[Scarecrow]

Jirka Lhotka, Francesco Salvi, Liudvikas Lazauskas
Department of Computer Science, EPFL, Switzerland

Abstract

Voynich manuscript is a 15th century document written in unknown language and script. 
Consequently, reading and transliterating it to electronic form is equivocal and leads to ambiguities. 
We present a way to disambiguate said uncertainties using word embeddings models, which achieves accuracies up to 86% on artificially corrupted texts of the same size.


You are not allowed to view links. Register or Login to view.