Ancient glass objects exhibit a wide range of compositions and styles, reflecting ancient societies' diverse production techniques and artistic traditions. The research of this paper focuses on the compositional analysis and identification of ancient glassware based on the K-means clustering algorithm, giving a specific classification of two types of glassware and verifying its rationality. Firstly, this paper constructs a decision tree classification model, and the importance of chemical composition characteristics of high-potassium glass, and lead-barium glass, both classified based on lead oxide. Using the K-means clustering algorithm to subclassify the two types of glass artifact samples, the elbow rule was used to determine the best subclasses of lead-barium glass into 5 subclasses and high-potassium glass into 3 subclasses, and the classification effect obtained is perfect, reaching 100%. This paper further refines the criteria and basis for subclass characteristics, conducive to simple and practical subdividing of types. The accuracy of the machine learning methods used in this paper is very high, which shows that the model chosen in this paper is very effective; secondly, this paper gives the specific classification standard for subclassification and reasonably verifies the accuracy of the standard. It is worth mentioning that the accuracy of the machine learning methods used in this paper is very high, which shows that the model this paper chose is very effective; secondly, this paper has given the specific classification standard for subclassification and reasonably verified the accuracy of the standard. This identification method not only ensures better accuracy but also helps to accelerate the identification speed, which is of guiding significance for the identification of ancient glass products in the future; this paper does not only stop at data analysis but also analyzes all aspects in combination with the chemical mechanism, which makes the model this paper built and the results this paper obtained more realistic value.

Yanrui Wang. Compositional Analysis and Identification of Ancient Glassware. Academic Journal of Materials & Chemistry (2024) Vol. 5, Issue 3: 37-42. https://doi.org/10.25236/AJMC.2024.050306.

[1] White, E., et al. Typological Classification of Egyptian Glass Artifacts Based on Morphological Features. [J]. Glass Studies, 2021(17):421-435.

[2] Lee, H., et al. Comparative Study of Chinese and Persian Glassware Through FTIR Spectroscopy. [J]. Glass Research, 2023(12):55-68. 

[3] Brown, A., et al. Trace Element Analysis of Hellenistic Glass Vessels Using Electron Probe Microanalysis. [J]. Archaeological Science, 2022(48):315-328.

[4] Smith, C., & Jones, D. Characterization of Roman Glass Beads by Fourier-Transform Infrared Spectroscopy. [J]. Archaeometry, 2020(35):82-95.

[5] Strugaj Gentiana, Herrmann Andreas, Rädlein Edda. AES and EDX surface analysis of weathered float glass exposed in different environmental conditions [J]. Journal of Non-Crystalline Solids, 2021, 572.