基于非对称视角的作者相似性探测  

Author Similarity Detection Based on Asymmetric Perspective

作　　者：席崇俊 刘志辉[1] 张均胜[1] Xi Chongjun;Liu Zhihui;Zhang Junsheng(Institute of Scientific and Technical Information of China,Beijing 100038,China)

机构地区：[1]中国科学技术信息研究所,北京100038

出　　处：《文献与数据学报》2022年第2期37-52,共16页Journal of Library and Data

摘　　要：[目的/意义]针对传统作者相似性探测中存在的问题,完善作者相似度算法,在梳理领域知识结构、挖掘作者合作关系以及科研社区发现等方面具有重要意义。[方法/过程]从非对称视角,将作者发文量、关键词使用频率、行文习惯等因素纳入考虑,提出一种融合作者研究强度和词关联度的作者相似度算法,先用关键词数据进行实证分析,然后将其应用于引文数据,进行计算并对比分析。[结果/结论]实验结果表明,本文提出的非对称视角下融入研究强度后的作者相似度算法可以更准确地反映作者之间的相似性,提高作者推荐精度;融入词关联度以及基于引文数据的计算可以放大作者间的相似度,对挖掘作者潜在合作关系具有一定帮助。[Purpose/significance]Aiming at the problems existing in the traditional author similarity detection,improving the author similarity algorithm is of great signifi cance in combing the domain knowledge structure,mining the author cooperation relationship and scientifi c research community discovery.[Method/process]From the perspective of asymmetry,taking into account the author’s document quantity,keyword use frequency,writing habits and other factors,an author similarity algorithm integrated with author research intensity and word relevance is proposed.Firstly,the keyword data is used for empirical analysis,and then applied to the citation data for calculation and comparative analysis.[Results/conclusion]The experimental results show that the author similarity algorithm integrated with research intensity from the asymmetric perspective proposed in this paper can more accurately reflect the similarity among authors and improve the accuracy of author recommendation.The integration of word relevance and the calculation based on citation data can enlarge the similarity among authors,which is helpful to mine the potential cooperative relationship of authors.

关 键 词：作者耦合关系 研究强度 词关联度非对称视角 

分 类 号：G350[文化科学—情报学]