Comparing the impact of subfields in scientific journals
AAPS/123-QED
Comparing the impact of subfields in scientific journals
Xiomara S. Q. Chac´on
Institute of Mathematics and Computer Science,University of S˜ao Paulo, S˜ao Carlos, SP, Brazil
Thiago C. Silva
Universidade Cat´olica de Bras´ılia, Distrito Federal, Brazil andDepartment of Computing and Mathematics,Faculty of Philosophy, Sciences, and Literatures in Ribeir˜ao Preto,Universidade de S˜ao Paulo, S˜ao Paulo, Brazil
Diego R. Amancio ∗ Institute of Mathematics and Computer Science, Department of Computer Science,University of S˜ao Paulo, S˜ao Carlos, SP, Brazil (Dated: August 7, 2020)
Abstract
The impact factor has been extensively used in the last years to assess journals visibility andprestige. While the impact factor is useful to compare journals, the specificities of subfields visibilityin journals are overlooked whenever visibility is measured only at the journal level. In this paper,we analyze the subfields visibility in a subset of over 450,000 Physics papers. We show that thevisibility of subfields is not regular in the considered dataset. In particular years, the variabilityin subfields impact factor in a journal reached 75% of the average subfields impact factor. Wealso found that the difference of subfields visibility in the same journal can be even higher thanthe difference of visibility between different journals. Our results show that subfields impact is animportant factor accounting for journals visibility. ∗ [email protected] a r X i v : . [ c s . D L ] A ug . INTRODUCTION In recent years, the visibility of scientific papers, authors, journals and conferences havebeen used as an important feature to quantify research relevance and impact [1]. The numberof citations has been an important quantity to gauge the visibility and quality, and for thisreason, many research impact measurements have been devised based on citation counts [2].At the author level, for example, the h-index has been widely used as a proxy to scientificrelevance [3], despite the many criticisms [4].Citations also plays an important role in evaluating journals research output. The prestigeof scientific journals is oftentimes measured via citation counts, among other factors [5].One important citation index for journals is the impact factor (IF), which essentially givesthe average number of citations received by papers in a journal in the last 2 years. Inmany cases, researchers use journals impact factor (and other journal attributes) to identifythe most relevant venue to disseminate their research. While the impact factor has beena disseminated index to measure visibility and relevance, it has been mostly used at thejournal level [6]. In this sense, the specificities of subfields in journals have been mostlyoverlooked. This means that in a high impact journal, some subfields might have a lowervisibility. Conversely, in a low impact journal, particular subfields might be more visiblethan the journal as a whole.In this paper, we investigate the behavior of subfields visibility in scientific journals. Weanalyze whether there is a significant difference of visibility for different subfields in the samevenue. Our analysis was performed in 450,000 Physics papers published by the
AmericanPhysical Society . Some interesting results have been found. First we found that there is aconsiderable variability in subfields impact in some cases. The variability in the subfieldsimpact factor can reach up to 75% of the average impact factor of subfields in the
PhysicalReview Letters (PRL) journal. In recent years, the variability reached 50% of PRL impactfactor. We also found that the difference in the visibility of subfields of the same journalmight be higher than difference in the visibility of journals. This results suggests that notonly the venue is an important factor to predict papers visibility, but also the subfield insidethe journal.The study of subfields impact is important because it provides a granular view of journalsimpact. This information could be used by authors to provide a more informed decision on2he choice of journals to publish. The analysis of subfields impact might also be usefulto understand the dynamics of journals visibility along time. A more refined visibilityinformation in journals could also assist editors in identifying promising subfields or subfieldsthat are no longer representative in terms of visibility.
II. RELATED WORKS
Many factors have been found to affect papers visibility. The total number of citationsmight not be influenced only by the inherent quality and innovation [7]. The total numberof citations received in the recent past might be an indicative of how many citations a paperwill receive in the future [8]. This process is referred to as preferential attachment in thenetwork science field [9, 10].The journal in which the scientific paper is published is important to establish the correctaudience for the paper, which may affect the future number of citations [7, 11, 12]. Anotherimportant feature that could affect papers visibility is the prestige of the journal. Often-times, prestige is gauged by visibility measurements – such as the impact factor, CiteScore,eigenfactor and influence score [13].Prestige at the author level plays an important role in the success of papers. Renownedauthors are naturally more visible and therefore tend to receive more citations [14]. Thepreferential attachment is also a relevant factor driving the dynamics of authors’ citations. Arecent model showed that a more reliable description of the citation curve of authors shouldtake into consideration only the citations accrued by authors in the last 12-24 months [15].Some additional factors have also been found to have an effect on the visibility of papers.This includes interdisciplinarity of the subject being approached, the number of tables,figures, references and some textual factors, such as the title length [16–20]. While many ofthese studies focus on the visibility of journals, papers and major fields, here we perform avisibility analysis at the subfield level. More specifically, we analyze the impact of differentsubfields inside journals. 3
II. METHODOLOGYA. Dataset and PACS classification
The
Physics and Astronomy Classification Scheme (PACS) is a hierarchical classificationof Physics and Astronomy scientific papers. A PACS code comprises 3 elements: a pair of twodigits separated by a dot. The digits are followed by two characters (letters or positive andnegative symbols). In the first part of the code, the first digit represents the main category ofthe paper and the second digit is the subfield inside the field specified by the first digit. Thelast characters in the code provide an even more specific characterization of subfields. Forinstance, the PACS code 05 . . − a refers to the following classification: “0” represents the“ General ” field, “05” denotes “
Statistical physics, thermodynamics, and nonlinear dynamicalsystems ”. Finally, the last part “-a” represents the “
Nonlinear dynamics and chaos ” subfield.In our work we focused our analysis at the third hierarchical level, which corresponds to thecode “05 .
45” in the previous example. This is an intermediary hierarchical level that allowsus to analyze subfields that are neither too general nor too specific. A list of all subfieldsmentioned in this paper is available in the Supplementary Information.The dataset we used for the current paper is the dataset provided by the
AmericanPhysical Society (APS). This dataset provides citations for the APS journals:
PhysicalReview Letters (PRL),
Review of Modern Physics (RMP) and
Physical Review A - E (PRA-E). We obtained the citation data for over 450,000 papers published in APS journals between1983 and 2016. While our analysis used only the citation data, journal name, PACS codeand publication date, the APS dataset also provides additional information, such as DOI,title, authors names and affiliations. For each journal and year, we considered a subfieldrelevant if at least 50 articles were published in that field in the last two years. B. Comparing groups of papers
In this paper we compare the impact of subfields. The subfields might belong to the samejournals or to different journals. The difference in the visibility of subfields was computedusing the so-called citation success index [21]. We decided to use this measurement becauseit provides a clear interpretation of the differences in the impact factor (and citation distri-bution) between any two groups of papers. The success index S is designed to quantitatively4ompare the success of two journals, but the same concept can be extended to compare anyset of papers. Given two set of papers, the reference ( r ) and target ( t ) sets, the success index S tr is defined as the probability that a randomly drawn article from the target group willreceive more citations than an article drawn from the reference group. S tr can be computeddirectly from the citation distribution of t and r [21]: S tr = ∞ (cid:88) c =0 (cid:32) P t ( c ) + p t ( c )2 (cid:33) p r ( c ) , (1)where P t ( c ) is the fraction of papers in t with more than c citations, and p r ( c ) is the fractionof papers in r that received c citations.A relationship between the citation success index and the impact factor can be derivedfrom the definition of impact factor and the definition of the success index in equation 1.The following equation can also be used to compute the citation success index: S tr = f − f /
21 + qρ − k (2)where ρ = I t /I r , is the ratio between impact factors of sets t and r , respectively; k = 1 . q is a normalization factor and f is the rate of uncited papers in r . It has been shown that f can be described by the following logistic function [21], which allows us to obtain q from q = 11 − f . (3)The computation of S tr from equation 2 can be simplified whenever f is low (which typicallyoccurs when I r >
10) or I t > I r . In this case, S tr = 11 + ρ − k . (4) IV. RESULTS AND DISCUSSION
In Section IV A, we analyze some subfields statistics in journals. In Section IV B, wecompare the visibility of subfields in the same journal. In Section IV C, subfields of distinct journals are compared.
A. Subfields statistics
We start our study by analyzing the evolution in the number of relevant subfields ofjournals. Overall, in most of the considered journals, the number of subfields increases along5he recent years (see blue curve in Figure 1). We also measured the number of subfields byconsidering the heterogeneity in the number of articles published in each subfield. To doso, we computed the diversity of subfields, a measurement that has been used in networkscience and other fields [22–25]. According to the diversity index, if all fields have the samesize, the diversity of subfields corresponds to the total number of subfields. Conversely, ifthe only a few subfields have most of the published articles, the diversity of subfields will bemuch smaller than the total number of subfields. In the diversity index, such a heterogeneityis measured via entropy [23].
Year1020304050 N u m b e r / D i v e r s i t y o f s u b f i e l d s PRD
Number of subfieldsDiversity of subfields (No. of papers)Diversity of subfields (No. of citations)
FIG. 1. Evolution in the number of PRD subfields (blue curve). The orange and green curvesrepresent the diversity of subfields considering the number of papers and number of citations inPRD subfields, respectively.
The diversity of subfields measured in terms of the number of articles published for eachsubfield is shown in Figure 1 (orange curve). Note that there is a heterogeneity in thenumber of papers in each subfield, since the of subfields diversity is much smaller than thetotal number of fields. In 2015, almost 60 subfields were found; however, the diversity pointsto effectively only 40 subfields. We also computed the diversity of subfields considering the6umber of citations received by subfields (rather than the number of published papers).For the PRD journal, Figure 1 shows that both diversity measurements are similar. Thissuggests that of number of citations received by subfields follows the number of publishedpapers.A different scenario can be observed for both PRE and PRL journals in recent years(result not shown). While the diversity curves for the number of papers and citations followthe same behavior, the diversity of subfields considering the number of papers is higher thanthe diversity measured via citations. For instance, in 2015, 115 subfields were identified inPRL. In this same year, the diversity of subfields in terms of the number of papers wasabout 90 subfields. Conversely, the diversity of subfields measured in terms of citations wasonly about 78 subfields. This results suggests that some subfields are more cited than othersand this difference cannot be explained only by subfields size in PRL. To further investigatethe differences in subfields visibility, in the next section we compare subfields in the samejournal using the citation success index.
B. Comparing subfields in the same journal
In order to analyze how the subfields visibility varies along time, we computed the yearlyimpact factor (IF) of each PACS code. In Figure 2(a), we show the evolution of averageimpact factor of PRL subfields. We also show the evolution of PRL impact factor. Becausewe used the APS dataset, we are limited to the citations received by APS journals. As aconsequence, impact factors might not be the same reported by
Clarivate Analytics [26].Similarly to other results in literature, this sampling does not affect the comparison ofjournals and subfields citation data [15, 21].The average subfield IF is consistent with the journal IF. This happens for all consideredjournals. We however observe a variability of subfields impact along time, meaning thatdifferent subfields are more (or less) visible than the journal as a whole. Such a variability isevident when analyzing the coefficient of variation of subfields IF. Figure 2(b) reveals that,in the most recent years, the typical deviation is roughly 50% of the average IF. An evenhigher heterogeneity of subfields impact occurred in 1990. In that year, a typical deviationof 75% of the average IF was observed. This result suggests that different subfields beingpublished by PRL might have different visibility. Other APS journals display a similar7ehavior, however the coefficients of variation of subfields impact are typically below 0.50(result not shown). I m p a c t F a c t o r PRLSubfields (average)1985 1990 1995 2000 2005 2010 2015Year0.30.40.50.60.70.8 C o e ff i c i e n t o f v a r i a t i o n FIG. 2. Evolution of the impact factor for subfields in the PRL journal. In (a), we show bothaverage and standard deviation of subfields impact factor. In (b), we show the coefficient ofvariation of subfields impact factor.
To further investigate how different is the impact of subfields inside a journal, we used thecitation success index (see Section III B). This measurement provides a clearer interpretationregarding the difference of visibility (i.e. impact factor) between two subfields. In otherwords, given subfields A and B , the success index S AB comparing them gives the probabilitythat a randomly drawn article from A will be more cited than an article drawn B (see SectionIII B).To understand the differences in visibility, for each journal, we measured the the successindex between all pairs of subfields in the same journal. Because S AB + S BA = 100%, inour analysis, for each pair A and B , we only considered the maximum between S AB and S BA . The results obtained for PRB is shown in Figure 3. For this particular journal, wenote that the median success index (comparing all pairs of subfields) varies between 55-57%.8his means that typically the impact factor of subfields in PRB are similar. However, inparticular cases, some subfields are much more visible than others. The highest values ofsuccess index is highlighted in the red curve. For comparison purposes, we also show thesuccess index obtained when comparing PRL and PRB (gray curve). Typically, the differenceof visibility between PRL and PRB is more significant than the difference in visibility ofPRB subfields. However, for particular pairs of subfields, the difference of visibility betweensubfields is more significant than the difference of impact between journals (PRL and PRB).In 2015, the success index comparing PACS 85.25 and 85.75 reached almost 85%, while thedifference between PRB and PRL was roughly 65%. Year505560657075808590 C i t a t i o n s u cc e ss i n d e x ( % ) PRB maxPRB vs PRLPRL vs PRB
FIG. 3. Evolution of success index comparing the impact of all pairs of PRB subfields. The redcurve represents the highest success index obtained for a given year. The gray curve represent thesuccess index resulting from the comparison of PRL and PRB. Typically, the impact differencebetween PRL and PRB is more significant than the impact difference between PRB subfields. Alist of PACS codes is shown in the Supplementary Information.
In Figure 4, we show the distribution of success indexes when comparing all pairs of9RE subfields (boxplots). In all considered years, most of the success indexes are below60%, revealing that there is no significant impact difference in PRE subfields for most ofthe considered pairs. However, as observed for PRB (see Figure 3), some pairs of subfieldsdisplay very distinct impact factors. In 2015, the comparison between PACS 89.20 and 85.75yielded a success index higher than 85%. This is much higher In recent years, we can seethat the typical internal impact difference is lower than the impact difference between PRAand PRE (see gray curve). Differently from Figure 3, in terms of visibility, it seems thechoice of subfield inside PRE is more important than the choice between PRA and PREin particular years. This is clear e.g. in 1995 and 1996, when more than 75% of all pairsof subfields were found to have a visibility difference higher than the one observed betweenPRA and PRE.
Year5055606570758085 C i t a t i o n s u cc e ss i n d e x ( % ) PRE maxPRE vs PRAPRA vs PRE
FIG. 4. Evolution of success index comparing the impact of all pairs of PRE subfields. The redcurve represents the highest success index obtained for a given year. The gray curve represents thesuccess index resulting from the comparison of PRA and PRE. A list of PACS codes is shown inthe Supplementary Information.
Year5055606570758085 C i t a t i o n s u cc e ss i n d e x ( % ) PRA maxPRA vs PRBPRB vs PRA
FIG. 5. Evolution of success index comparing the impact of all pairs of PRA subfields. The redcurve represents the highest success index obtained for a given year. The gray and blue curvesrepresent the success index resulting from the comparison of PRA and PRB impact. A list ofPACS codes is shown in the Supplementary Information. . Comparing subfields of different journals While in Figures 3–5 we focused our analysis on the comparison of subfields in the same journal, we now compare the impact factor of subfields in different journals. The comparisonbetween all pairs of PRA and PRE subfields is shown in Figure 6. One interesting patternobserved here is that the curve of success index comparing both journals follows the samebehavior of the median comparing pairs of subfields. Therefore, in general, the visibilitycomparison of PRA and PRE is compatible with the comparison of the respective subfieldsvisibility. However, particular subfields have very distinct visibility, as revealed by thedynamics of the red curve in Figure 6. Particularly, in 2015, a very large difference invisibility was found when comparing PACS 05.70 (from PRA) and 47.65 (from PRE). Notethat such a difference in visibility is much higher than the typical visibility difference betweenPRA and PRE.The relevance of comparing subfields (rather than journals) can be noted when comparingsubfields from PRC and PRB journals, as shown in Figure 7. From 1986 to 1994, thesuccess index comparing PRB and PRC is compatible with the median of the success indexcomparing the respective subfields. However, from 2002 to 2015, it is clear that PRC andPRB have very similar values of impact factor, as revealed by values of success index veryclose to 50% (see gray and blue curves). In this same period, however, the typical differencebetween subfields visibility was close to 60%. Once again, for particular subfields, thecitation success index reached values close to 85%. While in 2015 the journals have thesame impact factor, the subfields represented by PACS 12.38 and 61.46 were found to havea difference in impact yielding a success index close to 80%.All in all, the results presented in sections IV B and IV C revealed that subfields intra-and inter- journals might have very distinct visibility. This is an important result since itcould be used as an additional information in research policies. From scholars’ perspective,the quantification of subfields visibility could assist researchers in their career path decisions.Along with other field attributes, it could be of interest to know beforehand the potentialvisibility of subfields before efforts are made to learn and produce knowledge in the field.Another interesting conclusion arising from the obtained results concerns the comparisonof journals. Our results show that in some cases the direct comparison of journals mightnot give all the information relevant. Two journals with similar impact factors might have12
994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
Year505560657075808590 C i t a t i o n s u cc e ss i n d e x ( % ) PRA vs PRE maxPRA vs PREPRE vs PRA
FIG. 6. Evolution of success index comparing the impact of PRA and PRE subfields. The redcurve represents the highest success index obtained for a given year. The gray and blue curvesrepresent the success index resulting from the comparison of PRA and PRE impact. A list ofPACS codes is shown in the Supplementary Information. very distinct visibility values when subfields are compared. This information could be usede.g. to improve predictions and evaluations that use the impact factor (in combination withother established research impact indexes).
V. CONCLUSION
The analysis of subfields impact is relevant to provide a more detailed information ofthe factors affecting journals visibility. In this paper we studied the variability of subfieldsimpact in a subset of Physics journals published by the
American Physical Society . Theidentification of subfields was performed using the classification scheme provided by APS.One interesting result arising from our study is that the difference in the visibility of sub-13
993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
Year5055606570758085 C i t a t i o n s u cc e ss i n d e x ( % ) PRC vs PRB maxPRC vs PRBPRB vs PRC
FIG. 7. Evolution of success index comparing the impact of PRC and PRB subfields. The redcurve represents the highest success index obtained for a given year. The gray and blue curvesrepresent the success index resulting from the comparison of PRA and PRE impact. A list of PACcodes is shown in the Supplementary Information. fields in a same journal might be higher that the difference of visibility between journals.Altogether, our results suggest that subfields visibility in journals might be not uniformand this information could be used to better understand the components affecting journalsimpact.While we focused the subfields visibility analysis via PACS classification, this work couldbe extended by considering other notions of subfields [27]. For example, subfields could beidentified without any type of classification scheme. In particular, the identification of sub-fields could be performed via community detection in citation (or co-citation) networks [28].Subfields could also be identified using co-occurrence word networks obtained from titleand/or abstract [29–31]. Collaboration networks could also be used to detect subfields [32].Finally, we also intend to extend this study to analyze the variability of subfields visibility14n other major fields.
ACKNOWLEDGMENTS
DRA acknowledges financial support from S˜ao Paulo Research Foundation (FAPESP Grantno. 16/19069-9) and CNPq-Brazil (Grant no. 304026/2018-2). TCS gratefully acknowl-edges financial support from the CNPq foundation (Grant no. 408546/2018-2). XSQCacknowledges Capes-Brazil for sponsorship.15
UPPLEMENTARY INFORMATION: SUBLIST OF PACS
This Supplementary Information lists the code of the subfields mentioned in this manuscript.The first two digits corresponds to the first hierarchical levels. In this study we analyzedsubfields at the third hierarchical level. This hierarchical level comprises 3 digits.1.
Code 05 : statistical physics, thermodynamics, and nonlinear dynamical systems,subfield: thermodynamics (05.70).2.
Code 11 : General theory of fields and particles, with subfields: symmetry and con-servation laws (11.30).3.
Code 12 : specific theories and interaction models; particle systematics, with subfields:quantum electrodynamics (12.20), quantum chromodynamics (12.38).4.
Code 21 : nmuclear structure, with subfield: nuclear forces (21.30).5.
Code 25 : nuclear reactions: specific reactions, with subfields: relativistic heavy-ioncollisions (25.75).6.
Code 25 : nuclear astrophysics, with subfields: nuclear matter aspects of neutronstars (26.60).7.
Code 37 : mechanical control of atoms, molecules, and ions, with subfields: atoms,molecules, and ions in cavities (37.30).8.
Code 47 : fluid dynamics, with subfields: magneto-hydrodynamics and electro-hydrodynamics (47.65).9.
Code 67 : quantum fluids and solids, with subfields: ultracold gases, trapped gases(67.85).10.
Code 62 : Mechanical and acoustical properties of condensed matter, with subfield:mechanical properties of solids (62.20).11.
Code 64 : equations of state, phase equilibria, and phase transitions, with subfields:general studies of phase transitions (64.60), phase equilibria (64.75).12.
Code 67 : quantum fluids and solids, subfield: ultracold gases, trapped gases (67.85).163.
Code 74 : superconductivity, with subfields: cuprate superconductors (74.72), super-conducting films and low-dimensional structures (74.78).14.
Code 75 : magnetic properties and materials, with subfields: general theory andmodels of magnetic ordering (75.10), spin transport effects (75.76).15.
Code 85 : electronic and magnetic devices; microelectronics, with subfields: super-conducting devices (85.25), magnetoelectronics; spintronics: devices exploiting spinpolarized transport or integrated magnetic fields (85.75).16.
Code 87 : biological and medical physics, with subfields: biomolecules: types (87.14),subcellular structure and processes (87.16).17.
Code 89 : other areas of applied and interdisciplinary physics, subfield: interdisci-plinary applications of physics (89.20).17
1] L. Waltman, “A review of the literature on citation impact indicators,”
Journal of Informet-rics , vol. 10, no. 2, pp. 365–391, 2016.[2] S. Redner, “How popular is your paper? an empirical study of the citation distribution,”
The European Physical Journal B-Condensed Matter and Complex Systems , vol. 4, no. 2,pp. 131–134, 1998.[3] L. Bornmann and H.-D. Daniel, “What do we know about the h index?,”
Journal of theAmerican Society for Information Science and technology , vol. 58, no. 9, pp. 1381–1385, 2007.[4] L. Egghe, “An improvement of the h-index: The g-index,” ISSI, 2006.[5] W. Gl¨anzel and H. Moed, “Journal impact measures in bibliometric research,”
Scientometrics ,vol. 53, no. 2, pp. 171–193, 2002.[6] B. Alberts, “Impact factor distortions,” 2013.[7] L. Bornmann and L. Leydesdorff, “Does quality and content matter for citedness? a compar-ison with para-textual factors and over time,”
Journal of Informetrics , vol. 9, no. 3, pp. 419–429, 2015.[8] D. R. Amancio, O. N. Oliveira Jr, and L. F. Costa, “Three-feature model to reproduce thetopology of citation networks and the effects from authors’ visibility on their h-index,”
Journalof Informetrics , vol. 6, no. 3, pp. 427–434, 2012.[9] H. Jeong, Z. N´eda, and A.-L. Barab´asi, “Measuring preferential attachment in evolving net-works,”
EPL (Europhysics Letters) , vol. 61, no. 4, p. 567, 2003.[10] M. E. Newman, “Clustering and preferential attachment in growing networks,”
Physical reviewE , vol. 64, no. 2, p. 025102, 2001.[11] V. Larivi`ere and Y. Gingras, “The impact factor’s matthew effect: A natural experimentin bibliometrics,”
Journal of the American Society for Information Science and Technology
Scientometrics , vol. 84, no. 3, pp. 821–833, Journal of Informetrics , vol. 11, no. 1, pp. 307–323, 2017.[15] F. N. Silva, A. Tandon, D. R. Amancio, A. Flammini, F. Menczer, S. Milojevic, and S. Fortu-nato, “Recency predicts bursts in the evolution of author citations,” arXiv , vol. 1911.11926,2019.[16] L. Leydesdorff, “Betweenness centrality as an indicator of the interdisciplinarity of scientificjournals,”
Journal of the American Society for Information Science and Technology , vol. 58,no. 9, pp. 1303–1319, 2007.[17] F. Rostami, A. Mohammadpoorasl, and M. Hajizadeh, “The effect of characteristics of titleon citation rates of articles,”
Scientometrics , vol. 98, no. 3, pp. 2007–2010, 2014.[18] L. Leydesdorff and I. Rafols, “Indicators of the interdisciplinarity of journals: Diversity, cen-trality, and citations,”
Journal of Informetrics , vol. 5, no. 1, pp. 87–100, 2011.[19] N. Onodera and F. Yoshikane, “Factors affecting citation rates of research articles,”
Journalof the Association for Information Science and Technology , vol. 66, no. 4, pp. 739–764, 2015.[20] D. R. Amancio, M. G. V. Nunes, O. N. Oliveira Jr, and L. F. Costa, “Using complex networksconcepts to assess approaches for citations in scientific papers,”
Scientometrics , vol. 91, no. 3,pp. 827–842, 2012.[21] S. Milojevi´c, F. Radicchi, and J. Bar-Ilan, “Citation success index- an intuitive pair-wisejournal comparison metric,”
Journal of Informetrics , vol. 11, no. 1, pp. 223–231, 2017.[22] H. Tuomisto, “A consistent terminology for quantifying species diversity? yes, it does exist,”
Oecologia , vol. 164, no. 4, pp. 853–860, 2010.[23] E. A. Corrˆea Jr, F. N. Silva, L. F. Costa, and D. R. Amancio, “Patterns of authors contributionin scientific manuscripts,”
Journal of Informetrics , vol. 11, no. 2, pp. 498–510, 2017.[24] H. F. Arruda, F. N. Silva, L. F. Costa, and D. R. Amancio, “Knowledge acquisition: Acomplex networks approach,”
Information Sciences , vol. 421, pp. 154–166, 2017.[25] L. Jost, “Entropy and diversity,”
Oikos , vol. 113, no. 2, pp. 363–375, 2006.[26] https://clarivate.com/ .[27] S. Milojevic, “Practical method to reclassify web of science articles into unique subject cate-gories and broad disciplines,”
Quantitative Science Studies , vol. 0, no. 0, pp. 1–24, 2020.[28] F. N. Silva, D. R. Amancio, M. Bardosova, L. F. Costa, and O. N. Oliveira Jr, “Using network cience and text analytics to produce surveys in a scientific topic,” Journal of Informetrics ,vol. 10, no. 2, pp. 487–502, 2016.[29] N. Castro and M. Stella, “The multiplex structure of the mental lexicon influences picturenaming in people with aphasia,”
Journal of Complex Networks , vol. 7, no. 6, pp. 913–931,2019.[30] D. R. Amancio, F. N. Silva, and L. F. Costa, “Concentric network symmetry grasps authors’styles in word adjacency networks,”
EPL (Europhysics Letters) , vol. 110, no. 6, p. 68001, 2015.[31] M. Stella, S. De Nigris, A. Aloric, and C. S. Siew, “Forma mentis networks quantify crucialdifferences in stem perception between students and experts,”
PLoS ONE , vol. 14, no. 10,2019.[32] M. P. Viana, D. R. Amancio, and L. F. Costa, “On time-varying collaboration networks,”
Journal of Informetrics , vol. 7, no. 2, pp. 371–378, 2013., vol. 7, no. 2, pp. 371–378, 2013.