EAA2020: Abstract

The advent of Bayesian modelling in radiocarbon studies allowed archaeologists to include a wide set of chronological constraints into their radiocarbon models. Such modelling enabled the obtention of tighter chronological confidence intervals and allowed to automatically check whether the radiocarbon samples are in agreement with the prior Bayesian constraints. We propose an alternative approach, not limited to radiocarbon, where the archaeologist uses software to build complex chronological models featuring termini post & ante quem, duration bounds and several kinds of synchronisms between chronological entities. In this model, chronological ranges obtained through radiocarbon are treated as any other date-ranges, namely as the combination of a terminus post quem and a terminus ante quem, though in a deterministic fashion, without an associated probability. The set of encoded constraints is then automatically checked by a fast algorithm in order to ensure its consistency. If the model is found consistent, an algorithm computes the narrowest possible ranges for the dates and durations of all chronological entities. These use cases will be demonstrated using a new software called ChronoLog. The advantage of our approach lies in the high speed of the deterministic algorithms involved, which enable the users to interactively query models with hundreds of constraints and obtain instantaneous results. Its downside is that no probability is associated to either the input constraints or the output ranges: the latter are guaranteed to be optimal (i.e. as narrow as possible) provided the input constraints are correct. We will discuss the pros and cons of our approach with regard to radiocarbon dating and compare it to the more classical Bayesian approach to chronological modelling.

Chronology, Radiocarbon, Modelling, Formal methods, Computational archaeology

 Abstract (PDF)

Eythan Levy¹