Mining the characteristic patterns of glass artifacts based on multivariate statistical analysis
基于多元统计分析的玻璃文物特征规律的挖掘

摘要

古代玻璃是东西方经济文化交流的重要见证，其成分，类别上的差异反映东西方不同的工艺特色与地质特征。本文基于多元统计分析方法，从定性和定量两种视角，对不同类别的古代玻璃开展多层次的关联性分析，采取定性的卡方检验模型，发现其中类别对表面风化的影响较为显著；通过分析文物采样点化学成分的统计规律，使用定性的方差分析分别挖掘各化学成分与类别，表面风化之间关系，通过量化统计规律并据此预测风化点风化前的化学成分含量挖掘各个特征及其关系的统计规律，为古代玻璃样本的类型判断，成分分析提供统计依据。通过分析玻璃文物表面风化与三个定性变量之间关系可知其中二氧化硅，氧化钾等7种化学成分与类别和表面风化显著关联。

Abstract

Historic glass is an important witness to the economic and cultural exchange between the East and the West, and the differences in composition and type reflect the different technological and geological characteristics of the East and the West. Based on multivariate statistical analysis, this paper presents a multi-level correlation analysis of different categories of ancient glass from both qualitative and quantitative perspectives, and adopts a qualitative chi-square test model to find that category has a more significant effect on surface weathering. The statistical patterns were quantified and used to predict the pre-weathering chemical content of the weathering sites, and the statistical patterns of each feature and its relationship were explored to provide a statistical basis for determining the type and composition of ancient glass samples. By analysing the relationships between surface weathering and three qualitative variables, seven chemical constituents, including silica and potassium oxide, were found to be significantly associated with category and surface weathering.