摘要
随着数字技术发展,算法在不同领域的重要性日益提升,但少有论文探讨不同国家在算法研发工作中的表现。文章以自然语言处理领域为例,利用机器学习方法抽取该领域学术论文提及的算法,获得领域中的完整算法集合;收集所有算法的提出国家及相关信息,从算法数量和质量等维度分析不同国家的学术影响力。研究发现:在算法数量上,各国差异明显,中美是提出算法的主力军;在算法质量上,美国在综合被引上表现出色,各国平均被引差距较小,澳大利亚、乌克兰等排名明显提升;综合考虑算法数量和质量,将不同国家的学术影响力分为4种类型,从中发掘出数量、质量表现皆优的模范型国家,以及在单一维度影响力高的潜力型国家。文章为国家、机构、个人的学术影响力评价提供了新视角,验证了利用细粒度知识成果进行国家学术影响力评价的可能性,能为算法创新过程中各国之间的学术合作提供参考。
With the development of digital technology,algorithms have become increasingly important in various fields,but few papers have examined the performance of different countries in their algorithm development efforts.Taking the field of Natural Language Processing as an example,this article uses machine learning methods to extract the algorithms mentioned in the academic papers and obtain the complete set of algorithms.It collects the proposed countries and related information of all algorithms,and analyzes the academic influence of different countries in terms of algorithms,both quantitatively and qualitatively.The results show that there are obvious differences between countries given the number of algorithms,and that China and the United States are the main force in proposing algorithms;as for the quality of algorithms,the United States performs well in terms of comprehensive citations,the gap between the average citations of each country is small,and the rankings of Australia and the Ukraine are improved significantly.Taking into account the number and quality of algorithms,the academic influence of different countries can be divided into four types,from which exemplary countries with superior quantitative and qualitative performance can be identified,as well as potential countries with high influence in a single dimension.This paper provides a new perspective for evaluating the academic influence of countries,institutions,and individuals,and verifies the possibility of using fine-grained knowledge to evaluate a country’s academic influence,which provides a reference for the academic cooperation between countries in the process of algorithmic innovation.
作者
王玉琢
李晓婷
乔红
邢瀚文
章成志
WANG Yuzhuo;LI Xiaoting;QIAO Hong;XING Hanwen;ZHANG Chengzhi
出处
《图书馆论坛》
北大核心
2024年第5期54-66,共13页
Library Tribune
基金
国家自然科学基金项目“基于学术文献全文内容的细粒度算法实体抽取与评估研究”(项目编号:72074113)
江苏省社科基金重点项目“智能化驱动的学者细粒度画像构建研究”(项目编号:20TQA001)研究成果。
关键词
算法实体
学术影响力
影响力评估
国别差异
algorithm entities
academic influence
impact assessment
country differences