Composition Analysis and Identification of Ancient Glass Products Based on Random Forest Algorithm

This work investigates the machine learning-based identification method of ancient glass products. Exposure to burial environments can cause significant chemical exchange between the glass and environmental elements, leading to alterations in the composition and appearance of the artifacts and potentially influencing their correct classification. To address this issue, we have developed a statistical model to predict the original composition content of weathered glass artifacts based on composition and appearance data from known glass artifacts. In addition, we have employed the Random Forest algorithm to classify unknown glass samples. Our results demonstrate that the Random Forest algorithm is highly effective in distinguishing between different classes of glass artifacts. Through the analysis of composition content, we have observed substantial differences in the composition of glass artifacts of different types. Overall, our study contributes valuable insights into the identification of ancient glass products and presents a promising approach for predicting their original composition and classification.