Using Machine Learning to Study Archaeological Adhesives and Their Production Process
X. Yu (TU Delft - Mechanical Engineering)
Siddhant Kumar – Mentor (TU Delft - Team Sid Kumar)
M.W.E.M. Alfeld – Graduation committee member (TU Delft - Team Matthias Alfeld)
G.H.J. Langejans – Mentor (TU Delft - Team Joris Dik)
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
Adhesives have played a vital role throughout human history. Studying their composition and production methods offers insight into past technologies and helps reconstruct historical practices. This study focuses on the materials science analysis of Betula sp. (birch) bark tar, a widely used adhesive in prehistory times. By examining its molecular composition and production techniques, this research seeks to replicate ancient manufacturing methods using experimentally produced samples.
In this study, Gas Chromatography-Mass Spectrometry (GC/MS) was employed to analyze the chemical composition of the adhesives. To classify different production methods, machine learning techniques—including Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA), and K-Nearest Neighbors (KNN)—were applied. The results indicate that LDA successfully differentiates between production techniques, suggesting its potential for identifying variations in tar preparation. However, since this study is based on experimentally produced samples, its application to archaeological specimens requires further validation.