Use of Bibliometrics-Related Terms, their Evolution, and the Growth of Metrics in Science

Being bibliometricians, be it as a second choice, we wondered what to call our field. When I (RR) was introduced to the information sciences by Leo Egghe (around 1984), we used the term bibliometrics. Yet, when we learned more about the field, we preferred the term informetrics (since 1988), although we were not followed in this by the majority of colleagues. In the 21^st century, things became even ‘worse’ in the sense that for most outsiders, the term bibliometrics refers to the subdomain of indicator studies and their use for research evaluation. In my eyes, this is an unfortunate development as it focuses on a small part of ‘our field’. Again, by learning more about ‘our field’.^[1] I conclude that what I do (or at least like to do) is to study the science of science, my preferred term over the latest years.

Coming from another discipline, namely management, I (YF) got acquainted with the term bibliometrics once metrics have been more extensively used in the evaluation of research. More recently, with the use of social media, the term altmetrics has been adopted in marketing.

These personal reflections led us to the following research questions.

Q1. What is the relative use of the different metrics-related terms? Which term is used the most?

Q2. What is the relative contribution of documents related to bibliometrics (in the broad sense) compared to the whole universe of scientific documents? Is this relative contribution increasing?

We will show the evolution of the most-used terms referring to the field of bibliometrics (used here in a broad sense). As similar studies have been done in the past, see e.g.,^{[2, 3]} this paper can be considered a partial update. We point out, see further, that our approach is considerably different from.^[3]

Articles referring to the broad field of bibliometrics have been assigned to different categories and have made use of several keywords over the years: bibliometrics, scientometrics, informetrics, webometrics and webmetrics, altmetrics or social media metrics; in addition, a small number of articles refer to cybermetrics, naukometrics, imetrics, and librametrics. Recently, the term “quantitative science studies” has come to the fore, mainly because a new journal of this name has been founded.

We point out that the broadest term, namely “science of science” and the most specific, namely bibliometrics (originally referring to books and libraries), were introduced at about the same time: 1934-35. The term “science of science”, although well-known to Derek de Solla Price, had almost disappeared over the years, but during the latest two decades has gained a new life, be it in a slightly different context. The latter term, bibliometrics, introduced by Otlet,^[4] see also^[5] was practically unknown outside the francophone world and was re-introduced much later by Pritchard.^[6] The term naukometriya = scientometrics was introduced by Nalimov in the Russian literature in 1966^[7] and made popular first as a Russian term^[8] and later through its translation as scientometrics and the use of this term as the title of a scientific journal in 1978. For a history of these terms, we refer to.^[1]

Besides showing the evolution in the use of the “metrics” terms in absolute numbers, we also want to find out their relative use, answering, for instance, questions such as “Which term is the most-used metric term”, “Which term(s) show an increasing use compared to other terms” and most importantly “Is the relative number of publications using metric terms increasing or decreasing in the whole database”. We queried three international databases (WoS, Scopus, and Dimensions) corroborating our main conclusions.

This article is an extended version of a presentation made at the 2022 STI conference, held in Granada (Spain).^[9]

Data are collected during July 2022 from three databases: the Web of Science (WoS), Scopus, and Dimensions. We focus on WoS data and use the other two databases to confirm the results obtained via the WoS. Results were always restricted to the period before 2022. Additional data on reviewers’ requests, were collected during July 2023.

We collected data in the Web of Science Core Collection (WoS) by performing queries related to the following terms in the Web of Science (WoS):

bibliometr*

scientometr* OR naukometr*

informetr* OR infometr*

webmetr* OR webometr* OR cybermetr*

altmetr* OR “social media metr*”

For historical reasons, we also searched for librametr* but it turned out that this term is rarely used, and then mostly by Indian colleagues or by colleagues who try to cover all possible metrics-related terms, see e.g., the discipline with many names, as Maltseva and Batagelj^[10] call it. We also mention the terms patentometrics and knowmetrics (a Chinese term), two other metric terms not used in this investigation. We included the term naukometr* together with scientometr* but note that, in the whole WoS, this term has been used only 11 times.

Another term that is sometimes used is iMetrics. Yet, a search for “i-metric*” OR “imetric*” in the WoS yielded mostly articles in astronomy or physics dealing with Bianch (type-) I metrics. Restricting to imetric* yielded about 15 metric-related items, but even more other ones. For these reasons (small numbers and possible confusion), we do not include this term, promoted by Milojević and Leydesdorff,^[11] in our investigation.

We also searched for the term “science of science” because it precedes the formal introduction (or practically coincides with it, as in the case of bibliometrics) of the other terms, and also for the personal reason mentioned in the introduction.

Moreover, we collected the data for the union of the resulting sets related to these metrics, including the term science of science. We point out that it is not feasible to search for ‘metrics’ as this term is also used in computer science, astronomy, and several other fields. We further note that, on the one hand, we consider the term infometrics as an error (the term should not be used), but on the other hand, we consider the terms webometrics and webmetrics as equally valid, be it that webometrics is the original term,^[12] while personally, we prefer the term webmetrics.

These searches were restricted to the period [1955-2021] and were performed once for all fields, (ALL=); hence including e.g., the journal Scientometrics), once as a topic search (TS=), once as a title search (TI=), and once as a topic search, restricted (by ‘analyzing’) to the WoS category of Information Science and Library Science. The difference between an (ALL=)- search and a (TS=)-search is discussed in.^[13]

Similar to the queries in the WoS, we performed searches in Scopus. Here we searched under TITLE-ABS-KEY for the same terms as in the WoS, restricting to the period [1960, 2021]. Advanced search in Scopus makes it possible to find the union of the articles including one of the metric terms, in a similar way as in the WoS.

We used the free version of Dimensions and searched under title and abstract (but this includes keywords). We note that Dimensions does not support wildcards making queries more complicated (but e.g., bibliometric and bibliometrics yield the same result).

The following Table 1 shows the number of publications found by different queries in the WoS. In this table and further on in this article, metric terms are represented by the most-important notion (so scientometrics stands for the results of e.g., TS = (scientometr* OR naukometr*).

Table 1 shows that among the metrics terms bibliometrics is the most used, say most popular one, followed by scientometrics. Yet, because there exists a journal Scientometrics, and not a journal called Bibliometrics (at least not in the WoS) the difference between the use of these two terms is much smaller for ALL than for TS. The terms webmetrics (and variants), “science of science” and the Indian term librametrics are the least popular. Because of the extremely low numbers, we exclude the term ‘librametrics’ from further discussion. The fifth column, i.e., the ratio ALL/TS between the second (ALL=) and the third column (TS=), reflects the influence of the journals Scientometrics and Informetrics and of the biennial conferences of the ISSI (International Society for Informetrics and Scientometrics) on the use of these terms. Finally, the sixth column TS/TI(LIS) provides an indication of the relative importance of these metric terms in the field of Information and Library Science. Here we see that informetrics and webmetrics are the most typical terms (the least dispersed to other fields) for the LIS field. Bibliometrics and certainly “science of science” is, relatively speaking, the best-known or most-used outside our field. The ratio TS/TI, in the last column, is calculated to answer the question “If the term is used, is this then mentioned in the title?” Values (ratios) here are similar, except for the term webmetrics which is not so often used in the title. Finally, we note that the values on the last line are not the sum of the previous ones, because many publications use more than a single metric term.

We also collected the timelines for each (TS=) search. These are discussed further on. Remarkably, informetrics and webmetrics reached their top (in absolute numbers) a few years ago.

The use of the terms scientometrics (starting in 1976) and bibliometrics (starting in 1969, which is the year in which Pritchard,^[6] introduced the term, both show a huge increase in their absolute use, see Figures 1–2. As the curves for ALL=bibliometr* and TS=bibliometr* almost coincide, we only show one of the two. For the other terms, we will always show the results of the two queries, where the curve for (ALL=) is, by definition, always situated above the curve for (TS=). We recall that the term bibliometrics was introduced by Otlet,^[4] but this was largely forgotten by 1969, explaining why the term makes its first occurrence only in 1969. We note that the vertical axis of most figures refers to the number of publications, those showing a ratio being obvious exceptions.

The next figure (Figure 3), using a 3-year moving average, shows the relation between scientometrics and bibliometrics over time (starting in 1980) for (ALL=) and for (TS=). This yields a remarkable result. Because of the journal Scientometrics, the term scientometrics is in the beginning years more used than the term bibliometrics (ALL=)-data, but this is never the case for the results of a TS= query. Through (Proceedings of) the International Society for Scientometrics and Informetrics (ISSI), it is only since 2007 that bibliometrics became the more popular one. Yet, as a term used in documents (title, abstract, keywords) bibliometrics has always been the more popular one. Its relative use with respect to scientometrics has been increasing in the WoS, especially since the year 2000, leading to a decreasing trend line for the variable (use of scientometrics)/(use of bibliometrics). This is shown in a separate graph (Figure 4).

The next figure (Figure 5) shows the use of the term informetrics. Here we see a clear increasing trend. The first occurrence of this term was in 1979, the year in which the term has been proposed. The difference between the two curves is due to the Journal of Informetrics, and, leading to a jagged effect for the All= curve, the biennial conferences of the ISSI.

Figure 1:
Timeline for TS=bibliometr* in the WoS

Figure 2:
Timelines for ALL= (highest curve) and TS= (scientometr* OR naukometr*) in the WoS

Figure 3:
Ratio of the use of the term scientometrics versus the use of the term bibliometrics, ALL= (highest curve) and TS= (lower curve) results, WoS data

Figure 4:
Ratio of the use of the term scientometrics versus the use of the term bibliometrics, only (TS=) results, WoS data.

Figure 5:
Timeline for All= (highest curve) and TS = (informetr* OR infometr*), WoS data.

Figure 6:
Timeline for ALL= (highest curve) and TS= (webmetr* OR webometr* OR cybermetr*), WoS data.

The use of the term web(o)metrics begins in 1997, the year in which the term was introduced by Almind and Ingwersen.^[12] It shows an increasing trend until 2017-2018 but seems on the decline since then, see Figure 6.

Figure 7:
Timeline for ALL= (highest curve) and TS=(altmetr* OR “social media metr*”), WoS data

Figure 8:
Timeline for the term “science of science”, WoS data (ALL= leads to the higher curve)

Figure 9:
Timeline for the term bibliometrics in Scopus.

Figure 10:
Timeline for the term scientometrics in Scopus.

The first use of the notion of “altmetrics” occurred in 2005 as “social media metrics”. The term altmetrics itself was proposed by Priem et al.^[14] in 2010, but it is only since 2013 that the use of this term became increasingly popular, see Figure 7.

Finally, we see that the term ‘science of science’ occurred in this database (WoS since 1955) for the first time in 1956 and has a low but increasing trend, especially since about 2010 (Figure 8). Maybe this is related to the Science of Science and Innovation Policy (SciSIP) program launched by the US NSF in 2007. Since then, this program has been followed by the Science of Science: Discovery, Communication, and Impact (SoS: DCI) program. Recall though that the term “science of science” was already introduced in 1935^[15] in the Polish literature.

Results for Scopus and Dimensions are very similar to the WoS. We only show the timelines for bibliometrics and scientometrics and their ratio. For Scopus, we used the period [1960-2021]. Absolute and relative numbers are shown in Table 2 and discussed further on. Here, we show the timelines for bibliometrics (since 1969) and scientometrics (since 1976), as Figures 9 and 10.

Figure 11:
Ratio of the use of the term scientometrics versus the use of the term bibliometrics, Scopus data

Next, we compare the use of the terms bibliometrics and scientometrics in Scopus since 1980 (Figure 11). There is a decreasing trend over the whole period since 1980 (three-year averages), but contrary to the case of the WoS there is only a slight decline for the ratio (use of scientometrics)/(use of bibliometrics) over the latest years.

Figure 12:
Timeline for the term bibliometrics in Dimensions

Figure 13:
Timeline for the term scientometrics in Dimensions.

As for Scopus, absolute and relative numbers are shown in Table 2 and discussed further on. Again, we show the timelines for the terms bibliometrics (Figure 12) and scientometrics (Figure 13) and their ratio (Figure 14).

In the two other databases, we found a general decrease in the ratio (use of scientometrics)/(use of bibliometrics), but the behavior for the latest years was slightly different. Here, we see a confirmation of the results obtained for Scopus.

Figure 14:
Ratio of the use of the term scientometrics versus the use of the term bibliometrics in Dimensions

In the next table (Table 2) we show the absolute numbers of documents using the metric terms separately and as a percentage of the total, and this for the three international databases.

The percentage distribution of these terms in the three databases shows similar rankings with two dominant terms bibliometrics (around 80%) and scientometrics (between 16 and 19%), while the remaining terms group 10 to 12% of the uses. Dimensions and Scopus have the highest absolute use of metric terms, as a consequence of being the larger databases. A comparison between the three databases shows differences in the relative use of the ‘metrics’ terms. WoS has the highest percentages for all terms except bibliometrics and “science of science”. Scopus has the highest relative use for bibliometrics and Dimensions for “science of science”.

The next figure (Figure 15) shows our most important result, namely the number of items dealing with metrics divided by all items in the database (Scopus, Dimensions, and Web of Science Core Collection), multiplied by 10⁴. We consider the period since 1973. This graph shows that the relative increase in the use of “metric terms “started in earnest in the period between 2000 and 2005. The relative increase in the latest years, although showing a faster than exponential increase, is not as steep as for the absolute numbers. This proves that the increase shown in the previous figure is partly due to the growth of the databases as a whole. The highest curve refers to Dimensions, the second one to the WoS, and the lowest one to Scopus. Yet, all three show the same tendency.

This is a very important result showing the growth – with respect to all topics – in the relative importance of quantitative methods (metrics) to study researchers, research results, and science itself.

As a first indication of the diffusion over time of the metric terms in the WoS, we collected the number of WoS categories and Research Areas in which at least one of the metric terms occurred as the result of a (TS=)-query. This is a way to quantify knowledge export and diffusion from one academic field to another. We note though that the number of WoS categories has increased over time, see Table 3.

Figure 15:
Relative increase of the “metric” terms, period [1973-2021]. The vertical axis denotes the number of items dealing with metrics divided by all items in the database multiplied by 10^4. The upper curve refers to Scopus, the middle one to Dimensions, and the lower one to the WoS.

For each period, the leading WoS category and Research Area is Information Science Library Science, followed by Computer Science (for research areas). Computer Science Interdisciplinary Applications is most of the time the second WoS Category. Multidisciplinary Sciences slowly moves down the list, while Management and Business went up. In the latest period, Environmental Sciences jumped to third place, due to the journal Sustainability (a mega journal covering a wide range of topics).

The journal Scientometrics is by far the leading source for metrics-related topics, followed by the Proceedings of the conferences of the ISSI (under various names). More details about the leading journals (hence not including the Proceedings of the ISSI conferences) are shown in Table 4. We added results for American Documentation, Journal of the American Society for Information Science, Journal of the American Society for Information Science and Technology, and Journal of the Association for Information Science and Technology by the name of JASIST; similarly, we added results for Information Processing and Management and its precursor Information Storage and Retrieval. When journals are included in the WoS category Information Science and Library Science (and maybe another one) we indicate this by IS&LS in the third column. Otherwise, all categories are mentioned. The last column shows the year in which the journal was founded. Besides the top 15 journals, we also added a few IS&LS journals which might be of interest to the reader.

A large group of articles published in journals completely related to another field present applications of bibliometric analysis in those fields. They mainly use basic bibliometric techniques and have a title such as “Topic: a (bibliometric/scientometric) (analysis/approach)”. For some of these megajournals such as Sustainability the number of bibliometric-related publications is high in absolute terms, but very small seen as a percentage of the total number of their publications

In earlier publications, colleagues tried to define and delineate the difference between the metric terms, see e.g.,^[6] including their famous Figure 15 in,^[6] but nowadays we consider it futile to try to find out the difference between bibliometrics, scientometrics, and informetrics; and aren’t most metric investigations in some way web-related? For this reason, we do not try to formulate an exact definition for the different metric terms. Obviously, to cover all metric fields and come close to a full recall, one must perform a search for all terms.

In the WoS, we found five articles using the term scientometrics (but not naukometriya or naukometria), all published in 1976, before the launching of the journal Scientometrics (in 1978), among which two by Derek de Solla Price, published in the International Forum on Information and Documentation.

The following Figure 16 shows the timelines for the number of publications obtained by the search ALL= (scientometr* OR naukometr*) and the number of publications in the journal Scientometrics (WoS data). Clearly, in the earlier years, the two almost coincide, while there is a marked difference between them since about the year 2005. In later years the influence of the proceedings of the biennial conference of the International Society for Scientometrics and Informetrics (ISSI), published under different names, becomes very clear.

We performed a similar investigation for informetrics, see Figure 17. The first use of the term informetrics occurred in the year 1979, the exact year in which the term was proposed by Nacke^[17] and Blackert and Siegel.^[18] We notice the influence of the Journal of Informetrics, launched in 2007, and of the ISSI conferences.

Comparing our timelines with those shown in^[3] we first notice that Mejia et al. only show the part between 2011 and 2020. More importantly, according to their timelines in 2013, the use of the terms bibliometrics and scientometrics is almost equal (in the WoS Core Collection), while this does not show at all in our Figure 3. Checking, we found that Mejia et al. did not do a (TS=)-search, but an (ALL=)-search. This was not clear from their explanation. Moreover, a (TS=) refers to the use of the corresponding term by scientists in their articles, while an (ALL=) refers to the use of the metric terms over the whole database, including their use in journal titles, conferences, and books. Both points of view are valid of course. With a (TS=)- search one misses some implicit uses of the terms, as all articles published in journals such as Scientometrics and the Journal of Informetrics can be assumed to deal with metric-related topics. Yet, an article published in the Journal of Informetrics may deal exclusively with a webmetric or altmetric topic, or may only use the term bibliometrics in the body of the article, without mentioning it in the title, abstract, or keywords.

Figure 16:
The term scientometrics and the influence of the journal with the same name, WoS data, (ALL=) data

Figure 17:
The term informetrics and the influence of the journal with the same name. WoS data, (ALL=) data

The curves for bibliometrics and scientometrics, in the WoS as well as Scopus, have a clear “exponential-like” increasing trend (a best-fitting exponential function stays for recent data below the observed points). Yet, we know that such a trend cannot continue and at some point, the increase will slow down, leading to an S-curve see e.g., p. 298 in.^[19] We have, however, not yet reached the onset of an S-curve.

Figures 16 and 17 show how the inclusion of a specific technical term in the title of a prominent journal or leading conference of a field can have a significant impact on the frequency of use of the term as a keyword or topic. In the introduction, we mentioned the term “quantitative science studies” and a new journal with this name. Suggested by the editor we had a closer look at this term. Concretely, we performed the following search in the WoS on July 14, 2023:

TS= “quantitative science stud* “ OR TS= “quantitative studies of science”, resulting in 59 publications. Table 5 gives the corresponding time distribution.

The oldest use dates from 1983 when András Schubert started a regular feature in the journal Scientometrics, entitled “Quantitative studies of science – a current bibliography”, which explains the relatively high number of occurrences before 1990. Recent use occurs mainly in the journal “Quantitative Science Studies”, especially during its first year (2020).

When performing a bibliometric survey of a field it is important to assure that no “important” articles are missed. We first note that a search for the most popular term (in the WoS), namely ‘bibliometrics’ yielded 63% (ALL), 81% (TS), and 77% (TI) of all results (see Table 1). An interesting analysis, besides the results of individual searches, is the study of searches with a combination of keywords. Combining bibliometrics and scientometrics via an OR-search yielded, respectively 92%, 93%, and 92% of all results. Would that suffice for some purposes? Checking what we missed, we found for the (TS)-query, 22 articles with more than 100 citations, among which 16 related to altmetrics and social media metrics, 3 related to webmetrics, and 3 related to informetrics. So, we would miss important contributions related to the newer subfields.

Table 6 shows the combination of different keywords (two by two) by an OR-query in the WoS. Table 6 does the same for the results of an AND-query. We note that the combination of all five keywords (OR-query) delivers 25,764 publications.

At the individual level, bibliometrics reaches 81% of the total, scientometrics 19%, whereas altmetrics provides 5% of the total and informetrics and webometrics even lower amounts.

Combining terms via an AND-query turns out to be very selective (Table 7). Altmetrics articles seem more related to bibliometrics (as keywords) with 30% of altmetric articles also included in ‘bibliometrics’. This only holds for 10% when using scientometrics, for 4% when using webometrics, and for 2% when combining with informetrics.

Besides the question of what we obtain using certain queries, an important question is what we miss with the choice of one keyword with respect to another. For this reason, Table 8 presents combinations for the use of the search terms ‘bibliometrics’’ with the other search terms.

The second column shows the number of articles for each other term (denoted as number), with their percentage placed in the third column. The fourth column presents the number of articles selected with the OR function, with their percentage in the fifth column. The sixth column presents the number of articles selected with the AND search function, with their percentage added in the seventh column. Finally, the eighth column shows

the number of articles excluded (denoted as Excl) by the AND function; the ninth column gives the ratio of the excluded items for each search.

While 1,915 of the 5,016 ‘scientometrics’ articles are also retrieved by bibliometrics as a keyword, 3101 articles (or 62%) will be missed. This percentage, namely 62%, increases to 70% for informetrics and altmetrics, and to 78% for web(o)metrics.

The editor pointed out to us that our study is somewhat related in spirit to the early work of Derek Price^[20] on quantitative ways to understand and describe scientific activities and the growth of science.

In this paper, we took another look at the use of the metrics terms within the information sciences. An important caveat is the fact that the used databases only partially cover the Russian, Latin-American, Indian, or Chinese literature, biasing this study in favor of use in the West. The most important observation is the fact that the growth in the metric terms is higher than the growth of the databases themselves, indicating an increasing interest in bibliometrics (in the broad sense). We recall^[21] that e.g., the whole WoS database increases exponentially over time (but relatively close to linear growth).

We compared the use of metrics-related terms as registered in the WoS, in Scopus, and in Dimensions. Bibliometrics is by far the most used term to refer to quantitative studies of science. Generally speaking, the use of all metric terms shows a huge increase, and this in absolute as well as relative (with respect to the database) terms. The term bibliometrics has a slightly higher dominance in Scopus, occurring in more than 82% of all metrics-articles, than in WoS, where it occurs in about 80% of the (TS=)-searches, with Dimensions coming in between. If one is interested in the general scientific use of metrics terms, including in titles of journals or conference proceedings, then one must perform an (ALL=) query (using WoS terminology), leading to a considerable difference in the use of the terms scientometrics and informetrics. There is no overarching category that englobes all metric-related articles. While bibliometrics englobes some articles on informetrics, altmetrics, and webmetrics, this is less the case for the term scientometrics.

Of course, sometimes a term is used in another context. We found, for instance, the sentence “Philosophy is the science of sciences” in an old book review included in Dimensions. Similarly in Dimensions, if the original article has no keywords, then MeSH terms are added if they exist. Sometimes these MeSH terms were added some years later (after the publication of the article) explaining why a query for bibliometrics finds an article by Raisig,^[22] dealing with “statistical bibliography” before Pritchard (re-)introduced the term bibliometrics. This illustrates that the approach of adding MeSH terms makes perfectly good sense from an information retrieval point of view, but is, historically speaking, less accurate.

Our analysis shows the need for a better categorization system. This could require a post-publication review of the assignment of keywords to each article, possibly with first and second-order keywords. This same observation is valid for other scientific fields and subfields, with a variety of themes. Because science progresses in new subdomains and new themes such a categorization cannot be statically designed ad hoc but needs regular adaptations over time. Seen the imperfection of the categorization with imperfect keywords, a comprehensive bibliometric study should make use of a combination of related keywords to cover the whole field of study. This remark, of course, also holds for all studies related to a given field, not only to bibliometrics/scientometrics.

The term ‘bibliometrics’ has been adopted in all fields of science as the generic term for studies/applications of metrics through ‘bibliometric’ methods, whereas scientometrics, informetrics, and other related terms are more used by the specialists in the field, and have not followed the same dissemination towards other scientific fields. We note that all data referring to the introduction of metric terms can be found in.^[19]

A limitation of the present study is that we only studied the publication aspect, not the citation one. Hence, a more in-depth analysis should focus on the citation analysis of the metric fields. Is there a similar increase in the number of citations of metrics-related publications? Further, one could study if colleagues from certain countries, or in certain fields, have preferences for one of the terms.

The authors thank the anonymous reviewers and the editor for useful comments and suggestions.