基于LDA模型分类的中美欧人工智能发展比较研究

A Comparative Study on the Development of Artificial Intelligence in China,the United States and Europe Based on LDA Model Classification

近年来,人工智能技术快速发展,各国对人工智能的重视程度不断加深,人工智能的论文产出数量也在逐年渐递增。为了对各国人工智能研究水平进行比较,促进相关政策的制定,研究借鉴文本分析中LDA(Latent Dirichlet Allocation)模型对论文主题进行划分,并利用论文活动力指数、论文影响力指数测度各国在不同主题人工智能研究的发展水平。结果显示:美国、加拿大、中国处于人工智能研究的第一梯队,发表论文数量多,影响力较大;中国在2009—2020年之间论文发表数量迅速上升,同时在论文影响力、论文活动力方面得分较高,在细分领域方面,中国在计算机视觉领域、机器学习与底层算法领域有着较大优势。

In recent years,with the rapid development of artificial intelligence technology,various countries pay more attention to artificial intelligence,and the number of papers on artificial intelligence is also gradually increasing.In order to compare the research level of artificial intelligence in various countries and promote the formulation of relevant policies,this paper used the LDA(Latent Dirichlet Allocation)model in text analysis to divide the topics of the papers,and measured the development level of artificial intelligence research in different countries by using the activity index and impact index of papers.The results show that the United States,Canada and China are in the first echelon of artificial intelligence research,with a large number of published papers and great impact.The number of published papers in China increased rapidly from 2009 to 2020,and the scores in the impact and activity of papers were high.In the segmentation fields,China has great advantages in the fields of computer vision,machine learning and underlying algorithms.