What is it about?

In provenance analysis, identifying the origin of the archaeological artifacts plays a significant role. Usually, this problem is addressed by discovering natural groups in data measured with spectroscopic techniques. Then, principal component and classical partitioning cluster analysis are employed to reveal the groups that supposedly define the origin of the investigated artifacts. However, this work shows that maximizing the variance and searching for specific cluster structures can be misleading because it fails to discriminate clearly the different archaeological sources. In contrast, the new methodology reveals several acknowledged geological sources present in the materials through the exploitation of emergence and swarm intelligence without prior assumptions about the data structures. A combination of unsupervised and semi-supervised machine learning and chemometric is applied on samples of Mesoamerican geological sources and obsidian artifacts collected from the archaeological site of Xalasco in Mexico. The analysis of the artifacts showed a preference of Xalasco inhabitants to local obsidian deposits. The results show that this approach, in terms of robustness, is suitable for handling unbiased quantitative spectral analysis of archaeological materials revealing the natural groups of archaeological data.

Featured Image

Why is it important?

Using the raw spectra of obsidian artifacts, a PCA linear planar projection does not accurately discriminate between the existing archaeological sources and does not map the artifacts of unknown provenance to a known source. The topographic map of the generalized u-matrix, using the nonlinear DBS dimensionality reduction (projection method), visualizes the distance- and density-based structures of all the artifacts. Artifacts from the same obsidian source lie within the same valley. Different archaeological sources are divided by mountains, indicating larger distances and smaller densities between them. Besides a small subset (green), all artifacts collected from Xalasco were mapped to known sources.


Opens a new way to effectively process quantitative archaeological data. The proposed methodology combines techniques coming from chemometrics and data mining. XRF spectra of Mesoamerican obsidian sources were processed to predict the origin of archaeological artifacts collected from the site of Xalasco, Tlaxcala, Mexico. The provenance between the different geological sources was accurately discriminated in the results.

Dr denisse lorenia argote-espino
Instituto nacional de antropologia

Read the Original

This page is a summary of: Projection-Based Classification of Chemical Groups for Provenance Analysis of Archaeological Materials, IEEE Access, January 2020, Institute of Electrical & Electronics Engineers (IEEE), DOI: 10.1109/access.2020.3016244.
You can read the full text:



The following have contributed to this page