Abstract: China is a country of many different ethnic groups since genetic, geographical and culture formation of a different customs and ways of thinking. In order to study the eye movement behavior when the different ethnic groups observe things, Tobii eye tracker collection of the expression data that the different ethnic groups are used to solve Sudoku problem. The expression data from different ethnic groups are compressed into 16 dimensional data to construct the data sets; It is analyzed that the diversity of eye movement behavior data sets derive from Buyi, Mongolian, Korean, Dong and Han five ethnic respectively. Furthermore, a variety of machine learning method is explored to analyze the eye movement behavior data sets. The experimental results show that eye movement behavior diversity between different ethnic groups.

Key words: different ethnic groups, Tobii eye tracker, eye movement behavior, machine learning

摘要： 中国是一个多民族国家，不同民族由于遗传、地域和文化等因素形成了不同的风俗和思考方式。为了研究不同民族观察事物时的眼动行为模式，利用Tobii眼动仪采集不同民族在求解数独问题时眼动轨迹表达数据，将不同族系被试的表达数据压缩为16维数据向量来构建数据集；分析对比布依族、蒙古族、朝鲜族、侗族和汉族等5个民族眼动行为数据集的多样性，利用多种机器学习方法对得到的数据进行测试分析。实验结果表明，族系间眼动行为存在多样性，同时为进一步深入探索不同族系间眼动行为的多样性提供了依据。

关键词: 族系, Tobii眼动仪, 眼动行为, 机器学习