This paper focuses on the composition of ancient glass artifacts and the identification of types based on composition. Firstly, this study is to classify and summarize unknown ancient glass artifacts and perform sensitivity analysis, and secondly, to analyze the relationship between the chemical composition of different glass types and to derive the correlation differences in relation to the glass types. For problem one, a support vector machine classification model was developed, the sample data set was trained, the model obtained after training was saved, and then the data related to each chemical composition of the unknown type of ancient glass artifacts were imported, and finally their classification results were obtained, and a sensitivity analysis was performed in 10 groups for the highly significant indicator factor of silica content, when the silica content changed , while other factors remained constant, it was found that the classification results showed a great difference and the prediction probability changed significantly, so the established support vector machine classification model was considered to be highly sensitive. For the second problem, the internal relationship between the various chemical components of different types of ancient glass artifacts was analyzed, and here the Pearson correlation analysis was done between the different types of chemical components and each other, and several types were combined and the cross-sectional comparison was done by the heat diagram between them, and it was concluded that the correlation between barium oxide and copper oxide, phosphorus pentoxide and calcium oxide, barium oxide and sulfur dioxide in the type of lead barium The color effect of the thermogram is significant, while the correlation between phosphorus pentoxide and iron oxide, calcium oxide and potassium oxide, and strontium oxide and phosphorus pentoxide in the high potassium type is significant, and the color effect of the thermogram is significant.

Hui Xu. A study on the composition analysis and identification of ancient glass products based on SVM model. Academic Journal of Computing & Information Science (2022), Vol. 5, Issue 13: 89-95. https://doi.org/10.25236/AJCIS.2022.051314.

[1] Liu Yuanxin, Jiang Quan. Selection of evaluation method for weathering degree of flat glass [J]. China Building Materials Science and Technology,1992,1(6):29-33.

[2] Cristianini N, Taylor J S. An introduction to support vector machines and other kernel-based learning methods [M]. Translated by LI Guo-zheng, WANG Meng, ZENG Hua-jun. Beijing: Publishing House of Electronics Industry, 2004

[3] ZHANG Xue-gong. Statistical learning theory and support vector machines [J]. Acta Automatica Sinica, 2000, 26(1): 32-41.

[4] Vapnik, V. Statistical learning theory. 1998(Vol.3).New York, NY:Wiley, 1998: Chapter 10-11, pp. 401-492

[5] Zhou Zhihua. Machine Learning. Beijing: Tsinghua University Press, 2016: pp 121-139, 298-300

[6] Li Hang. Statistical Learning Methods. Beijing: Tsinghua University Press, 2012: Chapter 7, pp.95-135

[7] VAPNIK V N. The nature of statistical learning theory [M]. Translated by ZHANG Xue-gong. Beijing: Tsinghua University Press, 2000.

[8] VAPNIK V N. Statistical Learning Theory [M]. Translated by XU Jian-hua, ZHANG Xue-gong. Beijing: Publishing House of Electronics Industry, 2004.

[9] Mao Shisong, Wang Jinglong, Pu Xiaolong, et al. Advanced Mathematical Statistics (Second Edition) [M]. Beijing: Higher Education Press, 2006

[10] Byun H, Lee S W. Applications of support vector machines for pattern recognition: A survey [C]//International workshop on support vector machines. Springer, Berlin, Heidelberg, 2002: 213-236.