WHOSE TOOLS ARE THESE? AN ARTIFICIAL NEURAL NETWORK APPLIED TO THE CLASSIFICATION OF OLD KINGDOM EGYPTIAN CHISELS

This paper presents an attempt to apply advanced computational methods to a database of Old Kingdom Egyptian copper model tools. We examine a particular class of artefacts, chisels. A smaller dataset extracted from them was used to train several linear and non-linear classification models. All these models were able to classify the items according to their origin, the site or part of site where they were found. The origin of the chisels was set against a working hypothesis in an attempt to establish the provenance of some chisels, presumably ones coming from excavations by Hermann Junker at Giza, currently in the collection of the Kunsthistorisches Museum in Vienna. The classification model has corroborated the indications of other contextual information, and the tentative provenance of the assemblages in the Western Field at Giza is proposed. Another set of predictions was influenced by fragments of chisels, particularly those described in the database from Abusir South, which skewed the predictions of other fragmentary pieces towards this site.