Machine Learning Techniques for Classifying the Dual Writing System in Palaeohispanic Inscriptions: A Comparative Study of Neural and Non-neural Approaches

Resumen

The writing systems of the Celtiberian and Iberian languages -part of the Palaeohispanic languages- share a special characteristic: the dual system. This is crucial for the interpretation of the inscriptions because if it is present, some graphemes must be read differently. But it is not an evident property; therefore, this paper proposes the comparison of seven classification systems based on Machine Learning techniques, both quantitatively and qualitatively, to determine if they can be useful for historical linguists.