Artificial Neural Networks and ancient artefacts: justifications for a multiform integrated approach using PST and Auto-CM models

The integration of different approaches based on Artificial Neural Networks models has here been adopted to draw the guidelines of a map of a Mesopotamian administrative system. Two data sets concerning two different classes of findings have been contemporarily investigated using different models and procedures: a corpus of glyptic presentation scenes and group of administrative tablets from the archives of Umma. Both corpora are witnesses to the inner logics of late third millennium Mesopotamian state administration, and the investigations into them gave interesting contributions to the development of sound hypotheses for a general outline of the Ur III state bureaucratic culture. In fact, the results, obtained through different methodologies, show a large number of points of convergence, and the same features were recognized as "basic" both by Auto-CM and PST. In summary, through research on heterogeneous documents related to Ur III administrative communication, such as the relics of visual languages and traces of writing and sealing procedures, this work demonstrates how proper data mining techniques can partly reveal the very cultural background of some ancient centralized organizations and stimulate the development of new ways of considering the use and perception of those products