An innovative approach for risk assessment in archaeology based on machine learning. A Swiss case study. Quantitative approaches, spatial statistics and socioecological modelling

Castiello, Maria-Elena; Tonini, Marj (6 February 2019). An innovative approach for risk assessment in archaeology based on machine learning. A Swiss case study. Quantitative approaches, spatial statistics and socioecological modelling (Unpublished). In: International Colloquium on Digital Archaeology in Bern (DAB). University of Bern, Switzerland. 4th – 6th February 2019.

Preview

Text (abstract booklet) DAB2019_abstract_booklet.pdf - Published Version Available under License BORIS Standard License. Download (2MB) | Preview

In a world that is more and more complex and full of artificial, we strike to master the changes which continuously affect every single domain of our reality and especially challenge human intelligence and cognitive capacities. The study of artificial intelligence (AI) is being exhibited as a core strategy to meet growing demands of science and applications, to solve complex problems in various areas such as environmental science, finance, health-sector, etc. Machine learning (ML) is as a subfield of AI, mainly concerned with the development of techniques and algorithms that allow computers to learn from data. In an innovative way, this work intends to survey and demonstrate the effectiveness of bringing together traditional archaeological questions, such as the analysis of settlement patterns and past human behavior, with cutting edge technologies related to Machine Learning. Computations were carried out using R free software environment for statistical computing and graphics; data pre- and post-processing was performed in a GIS (Geographical Information System) environment. We provide a data-driven basis example of archaeological predictive modeling (APM) for the Canton of Zurich, Switzerland. Namely, a dataset of known archaeological sites of the Roman period was considered. The APM represents an automated decision making and probabilistic reasoning tool, relevant for archaeological risk assessment and cultural heritage management. We adopted Random Forest (RF) (Breiman, 2001), an ensemble ML algorithm based on decision trees. The model is capable of learning from data and make predictions starting from the acquired knowledge through the modelling of the hidden relationships between a set of input (i.e. geo-environmental features prone to influence site locations) and output variables (i.e. the archeological sites).As result, we obtained: 1) a ranking of geo-environmental features influencing the archeological site occurrence; 2) a map of probability expressing the likelihood of archaeological site presence, at different locations in a given landscape. These outputs become important not only to verify the reliability of the data, but also to stimulate experts in different ways: they are elicited to characterize the benefits and constraints of using such techniques and ultimately to think big about archaeological data.

Interest & Impact

Downloads

203 since deposited on 17 Jul 2019
47 in the past 12 months

Citations

5 Citations in Web of Science ®
6 Citations in Scopus

Search

in Google Scholar™

Services

Actions (login required)

Edit item

Actions (login required)

Edit item