(German Archaeological Institute | Johannes Gutenberg University Mainz | Rome, Italy)

Keywords: data mining, databases, archaeological reasoning, spatio-temporal patterns

Excavations are arguably one of the most important sources for archaeological data. Since the information gained from this data plays a significant role in the process of archaeological reasoning it was long argued, that the data itself should be accessible to evaluate the claims which were made based upon it. Although the development and use of databases as part of the excavation recording strategies has a long tradition in archaeological research, only in recent times, with an advancement in technology and more accessible database systems more excavations store their data as digital records in databases in a way that access to them could be provided easily.  By the means of integration, it is now theoretically possible to compare different excavations at the level of its records.
To achieve this, in my PhD Thesis, working with excavation data stored in iDAI.field, I study an exploratory and comparative approach to analyse the content of archaeological excavation databases. This approach is based on methods from data mining (in most cases in R and Python) with the goal to identify relevant, nontrivial patterns for the interpretation. Because of the evasive and incomplete nature of the archaeological record and the resulting uncertainty, the interaction between the archaeologist and the mostly exploratory algorithms like clustering, frequent pattern mining or outlier detection is considered essential. Avoiding blackboxing, identifying metadata which could be used for “Quellenkritik”, data visualization and the reproducibility of the achieved results are therefore an important part of this type of data analysis. My aim is, to offer an epistemological framework within which these methods could be applied.