摘要
由于语言的不同,中国作者在发表外文文献时很容易出现作者重名的问题,导致许多重名学者发表的学术文献无法很好地区分开来。针对这一问题,本文提出了一种基于图神经网络的姓名消歧算法,解决外文文献中的中国作者同名问题。首先,基于待消歧文献的属性特征及其关系构建异质学术关系网络,对文献进行表示学习;然后再进行聚类消歧。由于文献属性特征之间具有强关联性,本文在原有文献关系的基础上引入了消歧特征对来丰富节点关系类型。实验结果表明,本文提出算法的性能明显优于其他对比方法,有更好的消歧性能。
Due to the difference in languages,Chinese authors are more likely to share the same name to cause confusion when publishing foreign literature.It is a challenge to differentiate academic publications by multiple authors with the same name.To address the situation,this paper proposes the name disambiguation algorithm based on graph neural networks to solve the foreign literature′s Chinese author name ambiguity problem.Specifically,the paper first constructs a heterogeneous academic relational network based on the attribute features of the documents to be disambiguated and their relationships to learn the representations of the documents,and then perform cluster disambiguation.Due to the significant association between document attribute features,the paper introduces disambiguation feature pairs based on the original document relationship to enrich the node relationship types.The experimental results show that the performance of the proposed algorithm in this paper is significantly better than other comparative methods with better disambiguation performance.
作者
汤哲冲
方志坚
贾子杰
TANG Zhechong;FANG Zhijian;JIA Zijie(School of Information Science and Engineering,Zhejiang Sci-Tech University,Hangzhou 310018,China)
出处
《智能计算机与应用》
2024年第3期54-60,共7页
Intelligent Computer and Applications
基金
浙江省科学技术厅“领雁”研发攻关计划项目(2022C01220)。
关键词
姓名消歧
异质学术关系网络
消歧特征对
name disambiguation
heterogeneous academic relationship network
disambiguation feature pairs
