ASIAA Lab logo

Archaeology and AI: the use of Artificial Neural Networks

Medieval settlement is still today a fundamental topic for medieval archaeology, and in particular the settlement pattern of fortified villages in Tuscany is one of the most widely discussed argument. We know a lot about the organization and the spatial distribution of late medieval castles, but the causes of "incastellamento" are still unknown and also largely debated.

Standard spatial analyses can tell us a lot about ancient settlements and their relationships with natural resources, their roles in economy, the power mesh and so on. Much has been written on early and late medieval settlement in tuscany and a large number of quantitative methods and models were applied to this problematic. Unfortunately so complexes arguments can be only partially understood with “classical” statistic methods.

Artificial Neural Networks analysis on tuscany castles

A solution that can help us resolving this problem is offered by the recent development in Artificial Intelligence and in general by the Fuzzy Logic theory.

In this project our attention, due to the kind of the problem, is focused in particular on Artificial Neural Networks approach.

The initial skepticism on AI has finally been exceeded, and now these new methodologies are used in nearly all application, in the world of research and also in the businness reality.

Moreover, by using this method, the archaeologist is and remains the main and fundamental subject of the historical research; in this manner he does not endure the calculator, indeed his acquaintance is fundamental for the initial phase of the analisys, the network training. In fact is very important to give to the net the most corrected data as possible. In this sense artificial neural networks are similar to the children: if we teach them in the right way they will grow well, otherwise not.

Certainly castles were not only used like defensive fortress: their main importance resides in the central role that they had in the government of rural places. So each castle can be seen as an outcome of a background, but also as a feedback for future settlement development. The organization of medieval castles is not well known yet, but ANN can help us understanding how and why the "incastellamento" phenomenon grew up.

In this project ANN are used to help us estimating continuities and discontinuities between fortified villages in southern Tuscany between the XII and the XIII century and other settlements systems of other historical periods, both previous and successive. Site locations and quantitative grids representing environmental and ecological variables, like morphology or distances from natural resources, had been used to train networks. It is also possible to use ANN to measure differences in the medieval settlement system inside the Tuscany boundaries between the different sub-regions or to build a geographically-based classification of the settlements categories from late antiquity to the end of the middle ages in central Italy.

2003-2005 © webmaster | xhtml | css