A study on the classification of ancient glassware based on unsupervised learning

Abstract

In the process of weathering, ancient glass objects have changed their composition ratio due to the large exchange of internal elements with external environmental elements, which affects the correct judgment of their categories. This paper establishes a relevant model for the classification and research of glass artifacts. The relationship between the surface weathering of these glass artifacts and the type, decoration, and color of the glass is analyzed; the statistical pattern of the content of chemical components with and without weathering on the surface of the artifacts is analyzed in conjunction with the type of glass, and the content of their chemical components before weathering is predicted based on the data of the weathering points. To study the classification pattern of high potassium glass and lead-barium glass; to select the appropriate chemical composition for each category for sub-category classification, to give specific division methods and division results, and to analyze the reasonableness and sensitivity of the classification results. Identify the categories to which unknown categories of glass artifacts belong, and analyze the sensitivity of the classification results. To analyze the correlations between the chemical components of glass artifact samples of different categories and compare the differences in the correlations between the chemical component contents of different categories.