Bayesian Seriation as a Tool in Archaeology
Ulrich Halekoh, Werner Vach · 1999
The seriation of objects based on findings or traits accompanying them is a major task in archaeological research. An example is the chronological ordering of graves based on the gravegoods found therein. Correspondence analysis is a well established tool to solve this task but has only limited capability to show the accuracy of the provided seriation. We introduce a new approach based on a stochastic model and a Bayesian analysis which allows for a given dataset not only to determine an estimate for the underlying order, but also to asses its variability. This hints to subgroups of graves that are nearly exchangeable with respect to their relative chronological position and hence contributes to prevent over-interpretation. We also provide means to characterize traits that carry substantial chronological information and may be useful for the classification of new graves. Additionally, our approach may enable the archaeologist to incorporate prior information as sex-specific classifications of graves or their stratigraphic relationships.