Abstract
Ceramics are one of the commonest sources of archaeological information, yet their abundance often confounds documentation and analysis. This article presents a new method of documenting and analyzing ceramics that includes laser-aided profile measurement to capture ceramic shape and other information quickly and accurately, resulting in digital outputs suitable for both publication and morphometric analysis.
Linked software and database solutions enable unsupervised machine learning to cluster shapes based on similarity, eventually assisting typological analysis. Following an overview of current practices in ceramic recording and both standard and computational shape classification analyses, the new approach is discussed in full as a documentary and analytical tool.
A case study from the Middle and Late Bronze Age site of Kaymakçı in western Anatolia demonstrates the benefits of the recording method and helps show that a combination of automated and manual shape clustering techniques currently remains the best practice in ceramic shape classification.