Articles by Giuseppe Fedele

The advances of digital technology to the museum world have led to the development of computational tools for the classification of information as well as consultation of semantically correlated documents. The work presented here consists in an experiment, organizing the textual descriptions relative to iconographic works by means of SOM (Self-Organizing Maps), which represent the most common algorithm of artificial neural networks in the category of non-supervised learning, i.e. without the control and contribution of knowledge on the part of the human operator. The system produces a bi-dimensional map in which the words, represented graphically by means of semantically correlated nodes, are contiguous and form agglomerates. Therefore, the visit to a virtual museum containing works located in different sites can take place following pathways which are “conceptually-oriented”, independent of the learner’s cultural background.

A project for a Demotic Inscriptions on Ostraka Database is being carried out in collaboration between ILC/CNR (Pisa), the Department of Electronic Engineering (Calabria University) and the Department of the Ancient World History (Egyptological section, Pisa University). The aim of the project is to analyse the digital colour images of demotic texts on ostraka (Medinet Madi, in Fayyum region) with the aid of computational tools. The module described in the paper is a neural component able to learn the graphical features of each demotic symbol, which has been previously segmented in the images thanks to a semiautomatic procedure. A specific neural network tries to recognize the text written in the images linking the symbols segmented within the ostraka images database to the correspondent symbols available on a virtual keyboard. The graphical interface is particularly useful for teaching and research activities on this type of archaeological documentation.