Elsevier – OpenCitations blog

Additional 48 million citations in COCI, including references from IEEE

We announce the August 2022 release of COCI, the OpenCitations Index of Crossref open DOI-to-DOI citations, which is based on open references to works with DOIs within the Crossref dump dated August 2022. This new release extends COCI with more than 48 million additional citations, giving a total number of more than 1.36 billion DOI-to-DOI citation links.

This release includes citations from the articles published over the last four years by IEEE, whose bibliographic references were opened in June 2022.

A fundamental role in pushing the commercial publishers to open their citation data was played by Crossref’s recent announcement to change its reference distribution policy, by making all its metadata open.

Besides IEEE, COCI already includes the citation data derived from Elsevier (open via Crossref since December 2020) and from the last articles published by the American Chemical Society (whose references were opened in February 2021)

You can find more information about COCI in our open-access article

Ivan Heibi, Silvio Peroni & David Shotton (2019). Software review: COCI, the OpenCitations Index of Crossref open DOI-to-DOI citations. Scientometrics, 121 (2): 1213-1228. DOI: https://doi.org/10.1007/s11192-019-03217-6

Finally, just a reminder that the bibliographic and citation data in COCI:

- can be queried using the OpenCitations Indexes SPARQL endpoint;
- can be retrieved by using the COCI REST API;
- can be searched by using the OpenCitations Indexes Search Interface;
- are also available as dumps on Figshare in CSV, N-Triples, and Scholix; and
- can be freely re-used for any purpose.

Coverage of open citation data approaches parity with Web of Science and Scopus

Guest blog post by Alberto Martín-Martín, Facultad de Comunicación y Documentación, Universidad de Granada, Spain <albertomartin@ugr.es>

In this post, as a contribution to Open Access Week, Alberto Martín-Martín shares his comparative analysis of COCI and other sources of open citation data with those from subscription services, and comments on their relative coverage.

Comprehensive bibliographic metadata is essential for the development of effective understanding and analysis across all phases of the research workflow. Commercial actors have historically filled the role of infrastructure providers of bibliographic and citation data, but their choice of subscription-based business models and/or restrictive user licenses has significantly limited how users and other parties can access, build upon, and redistribute the information available on those platforms. Locking bibliographic and citation metadata behind these barriers is problematic, as it hinde r s innovation and is an obstacle to reproducibility.

Fortunately, the process of digital transformation that scientific communication is currently undergoing is providing us with the tools to get closer to the ideal of science as a public good. One of the most successful initiatives in this area is Crossr ef, arguably the single most critical piece of research metadata infrastructure currently in existence. I consider the best thing about it to be its commitment to openness. Not only is Crossref responsible for minting many of the DOIs that are assigned to academic publications, but it also publishes metadata about these publications (for over 120+ million records in their latest public data file) without imposing any access or reuse limitations.

Crossref metadata has already boosted innovation in a variety of academic-oriented tools. New discovery services such as Dimensions, The Lens, and Scilit all take advantage of Crossref metadata to keep their indexes up to date with the latest publications. The open-source reference manager Zotero is able to pull metadata associated with a given DOI from Crossref’s servers, providing an easy way to populate one’s personal reference collection that is more reliable than using Google Scholar. The Unpaywall database uses Crossref metadata (among other data sources) to keep track of which documents are Open Access, and this data is in turn used by Unsub, a service that helps libraries make more informed decisions about their journal subscriptions.

Historically, citation indexing has been a functionality available only from a few subscription-based data sources (most notably Web of Science and Scopus), or from free but largely restricted sources (e.g., Google Scholar). In recent years, however, commercial exclusivity over citation data has been waning. Digital publishing workflows make it easier for publishers to deposit the list of cited references along with the rest of the metadata when they register a new document in Crossref, and many are already doing it. Crossref’s policy is to make these lists of references publicly available by default, although publishers can elect to prevent their public release. From this, it follows that if most publishers deposited their reference lists in Crossref and consented to make them open, a comprehensive open citation index, one that is free of the restrictions present in traditional platforms, could be built.

The Initiative for Open Citations (I4OC) is an advocacy group that has been working since 2017 to achieve this precise goal, and it has already managed to convince a large number publishers (over two thousand) to open the references they deposit in CrossRef. In the first half of 2021, Elsevier, the American Chemical Society, and Wolters Kluwer joined this group, so that today all the major scholarly publishers now support I4OC and have open references at Crossref, with the exception of IEEE (the Institute of Electrical and Electronics Engineers). Thanks to the efforts of I4OC and the collaboration of publishers, 88% of the publications for which publishers have deposited references in CrossRef are now open. This has allowed organizations such as OpenCitations (one of the founding members of I4OC) to create a non-proprietary citation index using these data, namely COCI, the OpenCitations Index of Crossref open DOI-to-DOI citations. Other open citation indexes such as the NIH Open Citation Collection (NIH-OCC) and Refcat have also been recently released.

How do such open citation indexes compare to long-established indexes? In 2019, I set out with colleagues to analyze the coverage of citations contained within the most widely used academic bibliographic data sources (Web of Science, Scopus, and Google Scholar) to a selected corpus of 2,515 highly-cited English-language documents published in 2006 from 252 subject categories, and to compare this to the coverage provided by some of the more recent data sources (Microsoft Academic, Dimensions, and COCI). At that time, COCI was the smallest of the six indexes, containing only 28% of all citations. For comparison, Web of Science contained 52%, and Scopus contained 57%.

There are a number of reasons for those differences: first, at that point some of the larger commercial publishers including Elsevier, IEEE, and ACS, which routinely deposit references in Crossref, had not yet opened them. Second, many smaller publishers still do not deposit their reference lists in Crossref. Third, COCI only captures citation relationships between documents that have DOIs, thus missing citations to publications that lack them. Finally, while for our study data collection from all sources was carried out during May/June of 2019, COCI at that time had not been updated since November 2018, which increased its disadvantage when compared to other data sources with more frequent updates.

Since Elsevier is the largest academic publisher in the world, its recent opening of references at Crossref resulted in a significant increase in the total number of openly available Crossref references. The most recent version of COCI (dated 3 September 2021, and based on open references to works with DOIs within the Crossref dump dated August 2021) now contains both the processed references from Elsevier, and the references in the most recently published articles by ACS (the complete backfile of ACS references will appear in future versions of COCI).

Given these significant developments, how much has the picture changed? To find this out, I updated our 2019 analysis using the version of COCI released on September 3^rd 2021 and the NIH-OCC dataset released in the same month. To carry out a reasonably fair comparison while reusing the data extracted in 2019 from the other sources, I employed the same corpus of target documents, and only used citations in which the citing document was published before the end of June 2019. The intention was to learn how much the coverage of open citation data has grown as a result of the subsequent opening of reference lists in Crossref that were not public in 2019, and similar efforts. The results of this new comparison are published in [1].

The combination of COCI’s and NIH-OCC’s September 2021 releases contained more than 1.62 million citations to our sample corpus of documents from all areas, a 91% increase over the 0.85 million citations that we were able to recover in 2019 from COCI alone. Considering the citations available in all data sources, 53% of all citations are now available from these two open sources under CC0 waivers, up from the 28% we found in 2019. This coverage now surpasses the 52% found by Web of Science, and is much closer to the 54% found by Dimensions, and the 57% covered by Scopus. The relative overlap between COCI and the other data sources has also significantly increased: in 2019 COCI found 47% of the citations available in Web of Science, whereas now open citation data sources find 87% of the WoS citations. In the case of Scopus, in 2019 COCI found 44% of the citations available in Scopus: the percentage available from open sources has now increased to 81%. The number of citations found by COCI but not present in the other data sources has also widened slightly. These data are presented graphically in Figure 1.

Fig. 1. Percentage of citations found by each database, relative to all citations (first row), and relative to the number of citations found by the other databases (subsequent rows).

Where are these new citations coming from? Well, as we might expect, references from articles published in Elsevier journals comprise the lion’s share of the newly found citations in open data sources (close to half of all new citations), as shown in Figure 2. But there are also some IEEE citations here. This is because until recently reference lists from IEEE publications were available in the ‘limited’ Crossref category to members of Crossref Metadata Plus, a paid-for service that provides a few additional advantages over the free services Crossref provides. As a member of Crossref Metadata Plus, OpenCitations obtained these reference lists while they were available and included them in COCI. Subsequently, IEEE decided to make their references completely closed, explaining why references from more recent IEEE publications are not included in COCI.

Fig 2. The increases between 2019 and 2021 of citations indexed by open sources (COCI + NIH-OCC) from the articles of different publishers

There can be no doubt that open citation data is of benefit to the entire academic community. Thanks to COCI, NIH-OCC, and similar initiatives, and despite some setbacks, we are already witnessing how open infrastructure can help us develop models and practices that are better aligned with the opportunities that our current digital environment offers and the challenges that our society faces.

Conclusion: The coverage of citation data available under CC0 waivers from open sources is now comparable to that from subscription sources such as Web of Science and Scopus, offering a viable alternative upon which to base open and reproducible metrics of academic performance.

Reference:

[1] Martín-Martín, A., Thelwall, M., Orduna-Malea, E. et al. Google Scholar, Microsoft Academic, Scopus, Dimensions, Web of Science, and OpenCitations’ COCI: a multidisciplinary comparison of coverage via citations. Scientometrics 126, 871–906 (2021). https://doi.org/10.1007/s11192-020-03690-4

Crossing a significant threshold: more than one billion citations now available in COCI!

“The competitive benefits of closing access to citation data diminish with each new citation released to the public domain, but the benefits of open data remain. Going forward, citation data is almost completely public domain”.

With these words, from the article “A tipping point for open citations data” (July 15, 2021), Ian Hutchins celebrated the threshold crossing of one billion citations on public-domain databases in February 2021.

Now, a new significant milestone has been reached. We are enthusiastic to announce that COCI, the OpenCitations Index of Crossref open DOI-to-DOI citations has just been extended with 334 million additional citations. Its most recent release, the COCI July 2021 release, now contains a total of 1.09 billion DOI-to-DOI citation links derived from open references within Crossref,which includes the references of articles deposited or opened in Crossref between November 2020 and January 2021.

These numbers make us proud, and confirm the essential value of the Initiative for Open Citations (I4OC). Since 2018, the mission of I4OC has been to persuade publishers to provide open citation data by means of the Crossref platform. The I4OC untiring commitment has led the major academic publishers to a progressive change of heart regarding open citations, and the scholarly community to a deeper interest in this openness.

These factors contributed to the creation of COCI in 2018, the first open citation index created by OpenCitations, in which we applied the concept of citations as first-class data entities (Heibi I., Peroni S., Shotton D., 2019). Over the last three years, COCI has been extended in a series of releases, by harvesting citations mostly from Crossref data dumps, starting from an initial coverage of 300 million citations (First release).

A crucial event that preceded (and delayed!) this latest COCI release was Elsevier’s endorsement in the DORA Declaration on Research Assessment in December 2020, thereby making “reference lists for all articles published in Elsevier journals openly available via Crossref so they can be available for reuse. This means other important initiatives like I4OC can draw on this metadata”. As described in our previous post, Elsevier’s welcome commitment led to the opening of many previously closed references from its numerous academic journals submitted to Crossref. Now, after an extended period of data ingestion and processing, all these newly opened Elsevier references are available at OpenCitations within COCI.

Elsevier’s involvement has both an effective and a symbolical value. Even if publishing more than one billion citations is a thrilling achievement, and – as Hutchins wrote – we are now at a tipping point with regard to open citations data, this milestone is not the last stop. Together with the other organizations and projects that participate in the Initiative for Open Citations, we will keep claiming the urgency for the remaining academic publishers to join our cause, and sharing our values with the whole academic community to make all existing citations data freely open and accessible. Recalling what Dario Taraborelli wrote in the conclusion of his article “The citation graph is one of humankind’s most important intellectual achievements“, “the world is waiting for the citation graph to become a public good”.

Elsevier endorses DORA and opens its journal article reference lists

We congratulate and thank Elsevier, the world’s largest academic publisher, for endorsing the DORA Declaration on Research Assessment (https://sfdora.org/), thereby joining the hundreds of other publishers and scientific organizations which have endorsed DORA over the previous eight years, and also for making a commitment to open the references from all its journal articles submitted to Crossref. The text of Elsevier’s endorsement, dated 16th December 2020, is to be found at https://www.elsevier.com/connect/advancing-responsible-research-assessment, and includes the statement:

“We will make reference lists for all articles published in Elsevier journals openly available via Crossref so they can be available for reuse. This means other important initiatives like I4OC can draw on this metadata.”

Particular thanks are due to Kumsal Bayazit, Elsevier’s CEO, and to Andrew Plume, head of Elsevier’s International Center for the Study of Research (ICSR), for spearheading this change in stance on the part of Elsevier, which until this week has been alone among the major scholarly publishers in keeping its reference lists at Crossref closed, for which it has attracted much criticism from the academic community.

This change of heart on the part of Elsevier now means that by next spring, after Crossref has had a chance to implement this change in status over the large corpus of Elsevier journal metadata, the reference lists of articles in the vast majority of the world’s academic journals will be open, enabling such metadata to be used to enhance publication discovery and enable transparent research assessment. The I4OC web site and COCI, OpenCitations Index of Crossref open DOI-to-DOI citations, will reflect this change once it has happened.

The Institute of Electrical and Electronics Engineers (IEEE), the American Chemical Society (ACS), and the University of Chicago Press now stand alone as the only significant scholarly publishers who choose not to make their publication reference lists open.

The first issue of Quantitative Science Studies

The memorable date 20/02/2020 saw the publication by MIT Press of the first issue of Volume One of a new journal, Quantitative Science Studies (QSS), the official open access journal of the International Society for Scientometrics and Informetrics (ISSI). QSS’s Editor in Chief is Ludo Waltman (CWTS, University of Leiden, Netherlands), Vincent Larivière (Université de Montréal, Montreal, Quebec, Canada) and Staša Milojević (Indiana University Bloomington, Bloomington, Indiana, USA) are its Associate Editors, and it has a large and distinguished editorial board.

What makes the launch of this new journal remarkable is the story of how it came into being. In 2019, the entire editorial team of the Journal of Informetrics (JOI), a leading journal in this field published by Elsevier, resigned en masse and decided to start an alternative journal, QSS, both because of Elsevier’s position on open citations, and because, in their opinion, the financial model used by Elsevier violates the scientific ethos.

Reproducibility in the field of scientometrics requires scientific metadata that are both of high-quality and open, particularly those relating to bibliographic citations. The JOI editorial board was deeply concerned by the refusal of Elsevier to join almost all other large scholarly publishers in supporting the Initiative for Open Citations (I4OC). As we have previously reported on this blog, Elsevier is the largest contributor of bibliographic references to Crossref, but insists that these data should be kept closed.

Elsevier’s position, driven by commercial interests (since it sells access to citation data through Scopus), flies in the face of the scientific community’s clear move towards open science, with hundreds of scientometricians having signed an ISSI open letter urging scholarly publishers to support I4OC.

Science is a self-governing system, and the editorial team held the view that the ultimate responsibility for a scholarly journal should fall with the scientific community, who serve as the gatekeepers, producers, and consumers of scientific content.

The editorial team also believed Elsevier’s subscription fees to be excessive, and its article processing charges (APCs) for open access publishing to be unfairly high, thus limiting both those who can afford to read Elsevier journals and those who can afford to publish in them, so that publishing with Elsevier inevitably places major limits on scholarship, harming both science and society. It was for all these reasons that they forsook JOI and started QSS.

We at OpenCitations congratulate the editorial team for their courage in deciding to make this journal flip, and wish them, together with the ISSI and MIT Press, every success for this important new journal. We also commend the Technische Informationsbibliothek (TIB) – Leibniz Information Centre for Science and Technology and the Communication, Information, Media Centre (KIM) of the University of Konstanz, who, in collaboration with the Fair Open Access Alliance (FOAA), have generously agreed to cover APCs for the first three years of the QSS journal.

Openness of non-Elsevier references

For completeness, this post, also based on analyses performed by Daniel Ecer of eLife (d.ecer@elifesciences.org) on data he downloaded from Crossref in September 2017 (Ecer, 2017), complements the two preceding posts, and details the openness of references from scholarly publishers other than Elsevier.

The main conclusion is that, of the 650,093,489 references stored in Crossref from journal articles published by publishers other than Elsevier, 486,041,671 (74.76%) are open.

The detailed statistics derived from the Crossref data at the time of sampling relating to all publishers except Elsevier are as follows:

Number of works recorded at Crossref from publishers other than Elsevier

Crossref has records of 93,184,372 works with DOIs, of which 69,699,633 (74.80%) are journal articles and 23,484,739 (25.20%) are works that are not journal articles (i.e. book chapters, proceedings articles, datasets, etc.).

Of the 93,184,372 works, 76,795,932 (82.41%) were published by publishers other than Elsevier.

Of the 69,699,144 journal articles, 54,440,761 (78.11%) were in journals with publishers other than Elsevier.

Of the 23,484,739 works that are not journal articles, 22,355,171 (95.19%) were published by publishers other than Elsevier.

Numbers of non-Elsevier works with references

Of all 76,795,932 works with DOIs recorded in Crossref from publishers other than Elsevier, 27,609,963 (35.95%) have accompanying references and 49,185,969 (64.05%) lack references.

Of the 54,440,761 journal articles recorded in Crossref from publishers other than Elsevier, 23,459,805 (43.09%) have accompanying references, and 30,980,956 (56.91%) lack references.

Of the 22,355,171 works that are not journal articles recorded in Crossref from publishers other than Elsevier, 4,150,158 (18.56%) have accompanying references, and 18,205,013 (81.44%) lack references.

Number of non-Elsevier references at Crossref

Of the 1,075,133,743 references stored in Crossref from all works, 732,513,350 (68.13%) are from works published by publishers other than Elsevier.

Of the 956,050,193 references stored in Crossref from journal articles, 650,093,489 (68.00%) are from journals published by publishers other than Elsevier.

Of the 119,083,550 references stored in Crossref from works that are not journal articles, 82,419,861 (69.21%) are from works published by publishers other than Elsevier.

Average numbers of references per non-Elsevier work

The 732,513,350 non-Elsevier references stored in Crossref come from 27,609,963 works of all types with accompanying references, giving an average of 26.53 references per work.

650,093,489 non-Elsevier references come from 23,459,805 non-Elsevier journal articles with accompanying references, giving an average of 27.71 references per journal article.

82,419,861 non-Elsevier references come from 4,150,158 non-Elsevier works with accompanying references that are not journal articles, averaging 19.86 references per work.

Proportion of non-Elsevier works that have open references

Of the 27,598,963 non-Elsevier works of all type documented in Crossref that have accompanying references, 18,228,221 (66.05%) have open references.

Of the 23,459,805 non-Elsevier journal articles documented in Crossref that have accompanying references, 17,072,801 (72.77%) have open references.

Of the 4,139,158 non-Elsevier works documented in Crossref that are not journal articles and that have accompanying references, 1,155,420 (27.91%) have open references.

Proportion of non-Elsevier references that are open

Of the 732,513,350 references stored in Crossref from all works published by publishers other than Elsevier, 523,186,205 (71.42%) are open, and 209,327,145 (28.58%) are not open.

Of the 650,093,489 references stored in Crossref from journal articles published by publishers other than Elsevier, 486,041,671 (74.76%) are open, and 164,051,818 (25.24%) are not open.

Of the 82,419,861 references stored in Crossref from works published by publishers other than Elsevier that are not journal articles, 37,144,534 (45.07%) are open, and 45,275,327 (54.93%) are not open.

Proportion of references which are not open that are published by publishers other than Elsevier

Of the 551,932,682 references from all works stored at Crossref that are not open, 209,327,145 (37.93%) are from works published by publishers other than Elsevier.

Of the 470,008,522 references from journal articles stored at Crossref that are not open, 164,051,818 (34.90%) are from journal articles published by publishers other than Elsevier.

Of the 81,924,160 references from works that are not journal articles stored at Crossref that are not open, 45,275,327 (55.26%) are from works published by publishers other than Elsevier.

Details for all publishers combined, and for Elsevier separately, are given in the two previous posts.

Reference

Ecer, D. (2017). Crossref Data Notebook. Available at https://elifesci.org/crossref-data-notebook

Elsevier references dominate those that are not open at Crossref

Yesterday (November 23rd 2017) I was working with Daniel Ecer of eLife (d.ecer@elifesciences.org) to dig some hard facts out of the analyses he undertook on data he downloaded from Crossref in September 2017 (Ecer, 2017). Because of its dominant position in the scholarly publishing world, in this, the second of two related posts, I report the results for references from works published by Elsevier.

These show that, of all 956,050,193 references from journal articles stored at Crossref, 305,956,704 (32.00%) are from journal articles published by Elsevier, none of which are in the Crossref “Open” category, freely available for others to use.

Put another way, of the 470,008,522 references from journal articles stored at Crossref that are not open, 305,956,704 (65.10%) are from journals published by Elsevier.

On behalf of I4OC, I appeal to Elsevier to join the other major academic publishers and to submit and open all its references without delay.

The detailed statistics derived from the Crossref data at the time of sampling relating to Elsevier publications are as follows:

Number of Elsevier works recorded at Crossref

Of the 93,184,372 works of all types, 16,388,440 (17.59%) were published by Elsevier.

Of the 69,699,633 journal articles, 15,258,872 (21.89%) were published in Elsevier journals.

Of the 23,484,739 works that are not journal articles, 1,129,568 (4.81%) were published by Elsevier.

Numbers of Elsevier works with references

Of all 16,388,440 Elsevier works with DOIs recorded in Crossref, 10,835,273 (66.12%) have accompanying references, and 5,553,167 (33.88) lack references.

Of the 15,258,872 Elsevier journal articles recorded in Crossref, 10,212,958 (66.93%) have accompanying references, and 5,045,914 (33.07%) lack references.

Of the 1,129,568 Elsevier works that are not journal articles recorded in Crossref, 622,315 (55.09%) have accompanying references, and 507,253 (44.91%) lack references.

Number of Elsevier references at Crossref

Of the 1,075,133,743 references stored in Crossref from all works, 342,620,393 (31.87%) are from works published by Elsevier.

Of the 956,050,193 references stored in Crossref from journal articles, 305,956,704 (32.00%) are from journals published by Elsevier.

Of the 119,083,550 references stored in Crossref from works that are not journal articles, 36,663,689 (30.79%) are from works published by Elsevier.

Average numbers of references per Elsevier work

The 342,620,393 Elsevier references stored in Crossref come from 10,835,273 works of all types with accompanying references, giving an average of 31.62 references per work.

305,956,704 Elsevier references come from 10,212,958 Elsevier journal articles with accompanying references, giving an average of 29.96 references per journal article.

36,663,689 Elsevier references come from 622,315 Elsevier works with accompanying references that are not journal articles, averaging 58.92 references per work.

Proportion of Elsevier works that have open references

Of the 10,846,273 Elsevier works of all type documented in Crossref that have accompanying references, 417 (0.0038%) have open references.

Of the 10,212,958 Elsevier journal articles documented in Crossref that have accompanying references, none (0.0000%) have open references.

Of the 633,315 Elsevier works documented in Crossref that are not journal articles and that have accompanying references, 417 (0.0658%) have open references.

Proportion of Elsevier references that are open

Of the 342,620,393 references stored in Crossref from works of all types published by Elsevier, 14,856 (0.0043%) are open, and 342,605,537 (99.9957%) are closed.

Of the 305,956,704 references stored in Crossref from journal articles published by Elsevier, none (0.0000%) are open, 100% being closed.

Of the 36,663,689 references stored in Crossref from works published by Elsevier that are not journal articles, 14,856 (0.0405%) are open, and 36,648,833 (99.9595%) are closed.

Proportion of references which are not open that are published by Elsevier

Of the 551,932,682 references from all works stored at Crossref that are not open, 342,605,537 (62.07%) are from works published by Elsevier.

Of the 470,008,522 references from journal articles stored at Crossref that are not open, 305,956,704 (65.10%) are from journal articles published by Elsevier.

Of the 81,924,160 references from works that are not journal articles stored at Crossref that are not open, 36,648,833 (44.74%) are from works published by Elsevier.

Details for all publishers combined are given in the previous post, and those for all publishers other than Elsevier in the following post.

[Note: As a result of further calculations undertaken by Daniel Ecer on 27th November 2017, which are recorded in his updated Crossref Data Notebook (Ecer, 2017), the figures in this blog have been expanded to show the average number of references for Elsevier works submitting references. At the same time, very minor corrections have been made the total numbers of works in each category, which have not altered the percentages and main conclusions presented in this post.]

Reference

Ecer, D. (2017). Crossref data notebook. Available at https://elifesci.org/crossref-data-notebook

Milestone for I4OC – open references at Crossref exceed 50%

Yesterday (November 23rd 2017) I was working with Daniel Ecer of eLife (d.ecer@elifesciences.org) to dig some hard facts out of the analyses he undertook on data he downloaded from Crossref in September 2017 (Ecer, 2017). In this, the first of two related posts, I report the results for all publishers.

The analyses show that, of the 33,672,763 journal articles documented in Crossref that have accompanying references, 17,072,801 (50.70%) have open references, and of the 956,050,193 references from journal articles stored at Crossref, 486,041,671 (50.84%) are now classified as “Open”, and are freely available for third parties to download and use for any purpose.

This is a significant milestone for the Initiative for Open Citations (I4OC, https://i4oc.org/), which since early 2017 has been campaigning for scholarly publishers to open their reference lists, and a major gain for the world of open scholarship.

The academic community is deeply indebted to all those publishers whose references are now open (https://i4oc.org/#publishers), and to Crossref itself (https://www.crossref.org/), for making these references freely available, providing a tremendous resource for bibliometric analysis.

However, 51.7% of the journal articles recorded in Crossref lack accompanying references, and of the references that are submitted together with the metadata for the remaining journal articles, 49.16% are yet not open.

On behalf of I4OC, I strongly encourage those publishers who are not yet submitting references to Crossref with their article metadata to start to do so, and those other publishers who are submitting references but have not yet made them open to open them without delay, by sending a message requesting this to support@crossref.org.

The detailed statistics derived from the Crossref data at the time of sampling relating to all publishers are as follows:

Works with DOIs documented at Crossref

Numbers of works with references

Of all 93,184,372 works with DOIs recorded in Crossref, 38,445,236 (41.3%) have accompanying references and 54,739,136 (58.7%) lack references.

Of the 69,699,633 journal articles recorded in Crossref, 33,672,763 (48.3%) have accompanying references, and 36,026,381 (51.7%) lack references.

Of the 23,484,739 works that are not journal articles recorded in Crossref, 4,772,473 (20.32%) have accompanying references, and 18,709,979 (79.68%) lack references.

Numbers of references at Crossref

Crossref stores 1,075,133,743 references from all 93,184,372 works.

Of these references, 956,050,193 references (88.92%) are from journal articles, and 119,083,550 references (11.08%) are from works that are not journal articles.

Average numbers of references per work

The 1,075,133,743 references stored in Crossref come from 38,445,236 works of all types with accompanying references, giving an average of 27.97 references per work.

956,050,193 references come from 33,672,763 journal articles with accompanying references, giving an average of 28.39 references per journal article.

119,083,550 references come from 4,772,473 works with accompanying references that are not journal articles, averaging 24.95 references per work.

Proportion of works that have open references

Of the 38,445,236 works of all type documented in Crossref that have accompanying references, 18,228,638 (47.41%) have open references.

Of the 33,672,763 journal articles documented in Crossref that have accompanying references, 17,072,801 (50.70%) have open references.

Of the 4,772,473 works documented in Crossref that are not journal articles and that have accompanying references, 1,155,837 (24.22%) have open references.

Proportion of references that are open

Of the 1,075,133,743 references stored in Crossref from works of all types, 523,201,061 (48.66%) are open, and 551,932,682 (51.34%) are not open.

Of the 956,050,193 references stored in Crossref from journal articles, 486,041,671 (50.84%) are open, and 470,008,522 (49.16%) are not open.

Of the 119,083,550 references stored in Crossref from works that are not journal articles, 37,159,390 (31.20%) are open, and 81,924,160 (68.80%) are not open.

The majority of the references that are not yet open are from works published by Elsevier, as detailed in the next post.

[Note: As a result of further calculations undertaken by Daniel Ecer on 27th November 2017, which are recorded in his updated Crossref Data Notebook (Ecer, 2017), the figures in this blog have been expanded to show the average number of references for works submitting references. At the same time, very minor corrections have been made the total numbers of works in each category, which have not altered the percentages and main conclusions presented in this post.]

Reference

Ecer, D. (2017). Crossref Data Notebook. Available at https://elifesci.org/crossref-data-notebook