Publication boost in Web of Science journals and its effect on citation distributions

📝 Abstract

In this paper we show that the dramatic increase in the number of research articles indexed in the Web of Science database impacts the commonly observed distributions of citations within these articles. First, we document that the growing number of physics articles in recent years is due to existing journals publishing more and more papers rather than more new journals coming into being as it happens in computer science. And second, even though the references from the more recent papers generally cover a longer time span, the newer papers are cited more frequently than the older ones if the uneven paper growth is not corrected for. Nevertheless, despite this change in the distribution of citations, the citation behavior of scientists does not seem to have changed.

💡 Analysis

In this paper we show that the dramatic increase in the number of research articles indexed in the Web of Science database impacts the commonly observed distributions of citations within these articles. First, we document that the growing number of physics articles in recent years is due to existing journals publishing more and more papers rather than more new journals coming into being as it happens in computer science. And second, even though the references from the more recent papers generally cover a longer time span, the newer papers are cited more frequently than the older ones if the uneven paper growth is not corrected for. Nevertheless, despite this change in the distribution of citations, the citation behavior of scientists does not seem to have changed.

📄 Content

Publication boost in Web of Science journals and its effect on citation distributions Lovro Šubelja, * Dalibor Fialab a University of Ljubljana, Faculty of Computer and Information Science
Večna pot 113, 1000 Ljubljana, Slovenia b University of West Bohemia, Department of Computer Science and Engineering
Univerzitní 8, 30614 Plzeň, Czech Republic

• Corresponding author. Tel.: +386 (0)1 479 82 33. Email addresses: lovro.subelj@fri.uni-lj.si (L. Šubelj), dalfia@kiv.zcu.cz (D. Fiala). Abstract: In this paper we show that the dramatic increase in the number of research articles indexed in the Web of Science database impacts the commonly observed distributions of citations within these articles. First, we document that the growing number of physics articles in recent years is due to existing journals publishing more and more papers rather than more new journals coming into being as it happens in computer science. And second, even though the references from the more recent papers generally cover a longer time span, the newer papers are cited more frequently than the older ones if the uneven paper growth is not corrected for. Nevertheless, despite this change in the distribution of citations, the citation behavior of scientists does not seem to have changed. Keywords: Web of Science, publications, citations, references, distribution. Introduction
  It is well known that scientific communication has changed dramatically in recent decades. There has been a real publication boom with more and more papers published, indexed in databases, available online, and cited. All of this might have had some impact on the way research papers refer to each other and citation patterns come into existence. Although not necessarily all newly indexed publications in bibliographic databases are the result of new

research (Michels and Schmoch (2012) showed that half of the growth in the number of papers indexed in Web of Science from 2000 to 2008 was caused by the inclusion of previously existing journals), the growth of scientific production is undeniable and the question is whether this growth is accompanied by some novel trends in the citation patterns of research papers. In this study, we investigate this issue by analyzing two large Web of Science data sets consisting of computer science and physics journal articles, and conclude that the enormous increase in research publications alters commonly observed citation distributions. Nevertheless, when this growth is corrected for, the citation behavior of scientists appears not to change. Citation patterns of research papers and their change over the course of time were the concern of many previous studies. For instance, Egghe (2010) introduced a mathematical model of the aging of references. Larivière et al. (2009) documented that the age of cited references declined between 1900 and 2005, but, in contrast, Verstak et al. (2014) showed that more and more older papers are cited in current literature. As far as citation models are concerned, Eom and Fortunato (2011) modeled citation distributions in the papers from the American Physical Society (APS) journals and discovered the shifted power law function to best describe the citation patterns in the network under study. A combined exponential and power law citation model was proposed by Peterson et al. (2010). Newman (2014) describes a successful method for predicting the future impact of articles and another impact prediction model is discussed by Wang et al. (2013). Unlike the latter, Stegehuis et al. (2015) proposed a model that can be adopted soon after the publication of an article. Radicchi and Castellano (2011) inspected the citation distributions of all articles in APS journals between 1985 and 2009 in individual years and fields, and proposed rescaling factors that would enable impartial comparisons of citedness. A similar procedure was conducted by Radicchi et al. (2008) for papers from 20 different research disciplines. Also, Radicchi et al. (2009) analyzed the whole

collection of Physical Review papers (i.e. APS) from 1893 to 2006 and proposed a diffusion algorithm of scientific credit to rank authors by importance. Another study of this sort was that by Walker et al. (2007). Note that some of the above studies considered citation distributions, i.e. the proportion of citations papers published in a certain year receive in the subsequent years, whereas others focused on the proportion of papers that receive a certain number of citations over the same time period, i.e. the in-degree distributions of citation networks. Nevertheless, as we show below, our findings impact both types of studies. Data and Methods
In late 2014, we generated two citation networks of research articles: Computer Science (Web of Science papers categorized as “computer science”) with 492,124 nodes (articles) and 2,328,599 edges (citations) and Physics (Web of Science papers categoriz

This content is AI-processed based on ArXiv data.