摘要
本文提出了一种基于规则匹配和机器学习的论文作者名自动化消歧方法:首先基于人工构建的人名匹配规则确定候选作者,对于存在多个候选人的情况,基于论文的属性信息(例如合作者、标题、摘要、关键词和出版物名称等)提取特征,然后选取合适的机器学习算法进行消歧.实验效果表明K近邻和Softmax分类器较适合于论文作者名消歧任务;此外,将作者信息与论文的其他信息分开提取特征能够有效提高作者名消歧的准确性.
This paper proposes an automatic article author name disambiguation method based on rule matching and machine learning. For each article, the candidate authors are determined based on artificial constructed name matching rules firstly. For the cases of multiple candidates, features are extracted from the attribute information of the article, such as collaborators, title, abstract, key words and publication name, and then selected machine learning models are applied to author name disambiguating. The experimental results show that the K-nearest neighbor and Softmax classifier are more suitable for the author name disambiguation task than other models. In addition, extracting features of the authors information separatelycan from other information effectively improve the accuracy of the author namedisambiguation.
作者
邓可君
华凯
邓昌明
姜宁
袁玲
彭一明
张治坤
DENG Ke-Jun;HUA Kai;DENG Chang-Ming;JIANG Ning;YUAN Ling;PENG Yi-Ming;ZHANG Zhi-Kun(Computer Center, Peking University, Beijing 100871, China)
出处
《四川大学学报(自然科学版)》
CAS
CSCD
北大核心
2019年第2期241-245,共5页
Journal of Sichuan University(Natural Science Edition)
关键词
作者名消歧
机器学习
文本特征提取
Author name disambiguation
Machine learning
Text feature extraction