A Comprehensive Study on the Use of Multiple Classification Models for the Type Classification Problem of Unknown Artifacts

Abstract

The early glass was often made into bead-shaped ornaments in the West Asian and Egyptian regions and introduced to China. The ancient glass in China was made locally after absorbing its technology, and the appearance of glass products is similar to that of foreign ones, but the chemical composition is different. Therefore, it is a very meaningful task to analyze and identify the composition of ancient glass products. In this paper, based on the known chemical composition data of excavated artifacts, the statistical pattern of whether the artifacts are weathered or not is visualized by using frequency distribution charts, and the interference of random errors on the results is tested through a chi-square test. Subsequently, the two major categories of high potassium glass and lead-barium glass are classified, while the problem of classifying the types of unknown artifacts is solved. In this paper, a scientific, reasonable, and systematic system of glass-type classification is established, the prediction is carried out on this data set, and more accurate results are obtained with a correct rate of 90%, which proves that the model has a strong generalization and practical significance.