Many journals post accepted articles online before they are formally published in an issue. Early citation impact evidence for these articles could be helpful for timely research evaluation and to identify potentially important articles that quickly attract many citations. This article investigates whether Microsoft Academic can help with this task. For over 65,000 Scopus in-press articles from 2016 and 2017 across 26 fields, Microsoft Academic found 2-5 times as many citations as Scopus, depending on year and field. From manual checks of 1,122 Microsoft Academic citations not found in Scopus, Microsoft Academic's citation indexing was faster but not much wider than Scopus for journals. It achieved this by associating citations to preprints with their subsequent in-press versions and by extracting citations from in-press articles. In some fields its coverage of scholarly digital libraries, such as arXiv.org, was also an advantage. Thus, Microsoft Academic seems to be a more comprehensive automatic source of citation counts for in-press articles than Scopus.
Citation indicators derived from conventional scholarly databases, such as Scopus and the Web of Science (WoS), are often used for the impact assessment of published articles. They are rarely useful for recentlypublished and in-press articles, however, since their citation counts tend to be zero. The overall publication delay (the time between submission or acceptance and publication) also negatively influences citation indicators (Luwel & Moed, 1998;Yu, Wang, & Yu, 2005;Tort, Targino & Amaral, 2012;Shi, Rousseau, Yang, & Li, 2017). Traditional citation indexes seem to wait for articles to be formally published by journals before processing their references. For instance, on 15 October 2017 Scopus had indexed over 277,000 ''In-Press'' articles that had been published as "Online First" or similar in journals. Nevertheless, Scopus does not index or display the cited references of in-press articles until their final version is published in a journal issue (as of 20 October 2017 2 ). Hence, it seems likely that millions of citations from in-press articles are not included in any Scopus citation counts. WoS seems to wait for in-press articles to be published in an issue before reporting them. For instance, although on 15 October 2017 Scopus found 52 and 46 in-press articles in 2017 from Scientometrics and Journal of the Association for Information Science and Technology, respectively, none had been indexed in WoS.
Advance online publication increases citations to articles, Journal Impact Factors and Immediacy Index values (Alves-Silva et al., 2016;Al & Soydal, 2017;Echeverría, Stuart, & Cordón-García, 2017;Todorov & Glänzel, 1988). Many academic publishers provide early online access to their journal articles to minimize publication delays and perhaps to increase citation rates such as Springer (Online First), Wiley (Early View), Taylor & Francis (Latest Articles), and Nature Publishing Group (Advance Online Publication). Some authors deposit preprints or postprints (final drafts after peer review) of their articles to open access Some alternative sources of evidence have been proposed to identify the early intellectual impact of research, including article downloads (Kurtz et al., 2005;Brody, Harnad, & Carr, 2006;Bollen & Van de Sompel, 2008), Mendeley reader counts (Thelwall & Sud, 2016;Maflahi & Thelwall, 2017) and social web mentions (Thelwall, Haustein, Larivière, & Sugimoto, 2013;Zahedi, Costas, & Wouters, 2014). These all reflect types of use that are likely to appear before citations. Some, such as download counts, may reflect different degrees of interest or uses of academic research compared to citations (Kurtz & Bollen, 2010). Similarly, Mendeley reader counts partly reflect professional, teaching and educational uses (Mohammadi, Thelwall, & Kousha, 2016).
Several free scholarly websites, including Google Scholar, ResearchGate and Microsoft Academic, index or host preprint versions of articles that could be used for early citation impact assessment.
Google Scholar may be the largest index for the early citation impact of research because it generates higher citation counts than traditional citation databases. It is helpful in this regard by indexing different publishers and wider online sources, such as open access publications (Bar-Ilan, 2008;Khabsa & Giles, 2014;Halevi, Moed, & Bar-Ilan, 2017). For instance, the in-press article “Stochastic multicriteria decisionmaking approach based on SMAA-ELECTRE with extended gray numbers” published online first in the journal International Transactions in Operational Research on 7 February 2017 (DOI: 10.1111/itor.12380), had no Scopus citations by 25 September 2017 but had received eight Google Scholar citations from other recently-published journal articles (mostly online first). All eight citing journals found in Google Scholar were covered by Scopus and so indexing delays in Scopus were the reason for its missing citations. On 18 October, Scopus found two of the missing citations to the above article, confirming that indexing delays were the cause. These delays may be for technical (delays in accessing or processing publications) or quality control (waiting for the version of record) reasons. Despite the substantial Google Scholar coverage of scholarly publications and citations, it cannot be used for most research evaluations because it does not allow automatic data collection, which is a practical necessity for large scale analyses. The Publish or Perish software can extract Google Scholar citations and other citation impact indicators for individual papers, academics or journals, however (Harzing, 2007).
ResearchGate reports citation counts on its article profile pages by extracting citations from other publications uploaded to the site. It is therefore both a citation index and a digital repository. ResearchGate finds more citations to recently-published library and information science articles than both WoS and Scopus (Thelwall & Kousha, 2017). Neve
This content is AI-processed based on open access ArXiv data.
