A generative model of article citation networks of a subject from a large‑scale citation database

Livia Lin‑Hsuan Chang  · Frederick Kin Hing Phoa · Junji Nakano

Received: 1 July 2020 / Accepted: 8 May 2021 /

Published online: 4 July 2021 © Akadémiai Kiadó, Budapest, Hungary 2021

Abstract

  In this paper, we analyze the structure of the article citation network of a particular subject  obtained from the Web of Science (WoS) database. In specifc, we modify a model proposed in Caldarelli et al. (Phys Rev Lett 89(25):258702, 2002) and develop a generative  model for article citation networks in which an article receives citations based on a newly  defned property called “importance” introduced in this paper. Since the importance of an  article is quantitatively unmeasurable, we consider to use the in-degree of articles, which  is the number of citations that an article of interest is cited by other articles, as a surrogate quantity to describe an article’s importance. We simulate some in-degree distributions  to estimate the parameters of the tapered Pareto distribution. The generative model shows  good performance in the comparison between the generated data and data from the real  network, especially the citation network of recent years.

      在本文中，我们分析了从 Web of Science (WoS) 数据库中获得的特定主题的文章引用网络的结构。具体来说，我们修改了 Caldarelli 等人提出的模型。 (Phys Rev Lett 89(25):258702, 2002) 并为文章引用网络开发一个生成模型，在该模型中，文章根据本文中介绍的称为“重要性”的新定义属性接收引用。由于一个的重要性 文章在数量上是不可测量的，我们考虑使用文章的入度，即一篇感兴趣的文章被其他文章引用的次数，作为描述文章重要性的替代量。我们模拟一些入度分布来估计锥形帕累托分布的参数。生成模型在生成数据与真实数据的比较中表现出良好的性能 网络，尤其是近年来的引文网络。