Mapping Diamond Open Access in Belgium: Exploring the Differences Between Bibliometric and Manually Verified Approaches
In June 2026, we published the first version of the Belgian Diamond Open Access Landscape dataset, an openly available dataset documenting Belgian Diamond Open Access (OA) and delayed OA journals.[1] As the mapping exercise is conceived as an ongoing effort, we continue to identify, verify, and document additional journals. These efforts have already resulted in the publication of a second version of the dataset.[2] The most important changes concern two targeted review exercises. First, we systematically compared our dataset with the ALMASI Diamond OA dataset, a large-scale bibliometric dataset of Diamond OA journals in Africa, Europe, and Latin America.[3] This comparison allowed us to identify a number of Belgian journals that met our inclusion criteria but had been missed in the first version of our dataset. At the same time, the comparison provided valuable insights into the different outcomes that are produced by bibliometric and manually verified approaches. The results of this comparison, together with a detailed discussion of the observed differences between both datasets, are presented later in this article. Second, we conducted a systematic review of the Diamond OA publishing activities of Belgian university presses, as well as two Belgian-based scholarly publishers, Brepols and Peeters Publishers. Both exercises further strengthened the coverage of the Belgian scholarly publishing landscape.
While the comparison with the ALMASI dataset initially served a practical purpose, namely identifying additional journals for inclusion in version 2 of the Belgian Diamond OA Landscape dataset, it also generated a number of broader insights, which may be of interest to others working on Diamond OA monitoring and mapping. Before turning to the results, however, it is useful to give a brief overview of the methodologies used by both datasets. The ALMASI Diamond OA dataset was constructed by combining metadata from ROAD[4] and OpenAlex[5] to create a base list of OA journals. Journals that charge Article Processing Charges (APCs) were subsequently identified and excluded through cross-referencing with OpenAPC,[6] DOAJ,[7] and OpenAlex metadata. The resulting dataset was further validated and enriched through the integration of established Diamond OA datasets, national journal portals, and targeted manual verification in a small number of individual countries. Where information on publication fees was not explicitly available, journals were assumed to be Diamond OA and retained in the dataset. The final dataset contains 20.355 Diamond OA journals from Africa, Europe, and Latin America.[8]
In contrast, the Belgian Diamond OA Landscape dataset was built through a process of systematic manual verification and contextual interpretation at the level of individual journals. Journals were identified through a broad range of sources, including DOAJ, DDH,[9] OpenEdition Journals,[10] institutional publishing platforms, national funding programs, and existing Diamond OA inventories.[11] Each journal was subsequently reviewed individually, with the journal’s own website serving as the primary source for assessing journal characteristics and publication models. Candidate journals were then evaluated against a number of explicit inclusion criteria: (1) only active journals are included, meaning that the journal has published at least one issue within the previous five years,[12] (2) a journal is considered Belgian only when the publisher or publishing entity is legally based in Belgium, and (3) journals are classified as Diamond OA when no publication fees are charged to authors for unconditional open publishing.[13] In addition, the dataset separately documents Belgian delayed OA journals, allowing both publication models to be studied alongside one another while maintaining a clear distinction between them.[14]
The comparison between both datasets provides an opportunity to examine not only the completeness of both datasets but also the kinds of journals that tend to remain invisible under different methodological approaches. In particular, it allows us to assess the extent to which national, manually verified mapping exercises can complement international bibliometric initiatives, especially in a context where many Diamond OA journals remain only partially represented in major international indexes and databases. The relevance of such verification is also acknowledged by the authors of the ALMASI dataset, who note that bibliometric metadata are not always up to date and may contain inaccuracies regarding journals’ OA policies.[15] Similar concerns have been raised in studies of national Diamond OA landscapes outside Belgium, where it has been shown that the journals identified can vary depending on the criteria and data sources used.[16]
As said above, the first outcome of this comparison was the identification of additional journals for inclusion in the Belgian Diamond OA Landscape dataset. As a result, 63 journals are shared by both datasets, whereas 92 journals are unique to the ALMASI dataset and 99 are unique to the Belgian Diamond OA Landscape dataset. We also identified one duplicate record in the ALMASI dataset. The analysis therefore uses 155 unique journals rather than the 156 records reported in the original dataset.[17] These figures indicate that, despite a substantial overlap, both datasets also contain a considerable number of unique titles, which calls for a closer examination of the reasons for this divergence.

Figure 1. Comparison of journal coverage in the ALMASI dataset and the Belgian Diamond OA Landscape dataset.
Of the 92 journals that were not included in the Belgian Diamond OA Landscape dataset, the largest group consists of Hybrid OA journals (n = 49). Another substantial group comprises discontinued journals (n = 16), indicating that they are no longer active at the time of verification. Furthermore, 13 journals were excluded because they operate under a Gold OA model with Article Processing Charges (APCs), while for 9 journals no explicit evidence could be identified that all content would become openly accessible after an embargo period. A smaller number of journals were excluded due to the absence of a (current) Belgian connection (n = 5), because they provided paywalled access to their content (n = 3), or because they followed a Subscribe to Open model (n = 1).[18]

Figure 2. Number of journals excluded from the Belgian Diamond OA Landscape dataset by exclusion criterion.
Several explanations may account for the discrepancies between both datasets. One possibility is that journal information changed between the creation of both datasets. However, given that the ALMASI dataset was published on 11 December 2025 and the first version of the Belgian Diamond OA Landscape dataset was completed on 23 June 2026, this appears relatively unlikely, as the interval between both datasets is limited. A more plausible explanation is that some journals were incorrectly classified in the bibliometric sources on which the ALMASI methodology relies. Inaccurate, incomplete, or outdated metadata concerning publication models, APC policies, journal activity status, and publisher information most likely explains why a number of journals identified as Diamond OA by ALMASI did not meet the inclusion criteria applied in the Belgian Diamond OA Landscape study.
If we zoom in on the Hybrid OA journals, an interesting observation is that more than one-third of the journals identified in the ALMASI dataset (57 out of 155) are published by Brepols and Peeters Publishers, two publishers based in the Flemish region of Belgium and focused on the humanities. The strong presence of these two commercial academic publishers largely explains the high number of journals that were ultimately classified as Hybrid OA rather than Diamond OA. A likely explanation is that the publication models of these journals are difficult to identify through bibliographic metadata alone. These journals contain a mixture of open and restricted content, while bibliographic databases do not always capture the distinction between access conditions and publication models in a sufficiently granular manner.[19] What is also striking is that several journals published by Brepols and Peeters Publishers that were identified as Diamond OA in the Belgian Diamond OA Landscape study were absent from the ALMASI dataset. This suggests that bibliographic metadata not only leads to the inclusion of journals that do not meet the Diamond OA criteria, but also results in the omission of journals that do.
Not all differences between the two datasets can be attributed solely to inaccuracies or limitations in bibliographic metadata. The category “Embargo not explicitly stated” reflects a methodological difference in how delayed OA journals are identified and classified, rather than a disagreement about the OA status of these journals. No information could be identified regarding how delayed OA journals were handled when compiling the ALMASI dataset. In our study, delayed OA journals were explicitly included as a separate category, provided that the journal clearly stated that all content would become openly accessible after a defined embargo period. This requirement was adopted as a deliberate inclusion criterion in order to establish a clear and consistently applicable boundary between journals that could be verified as delayed OA and those for which insufficient evidence was available. Although delayed OA journals were also present in the ALMASI dataset, they were not identified as a distinct category. This suggests that delayed OA journals were captured by their methodology, but without the additional verification criteria that we applied.
Although this concerns only a single journal in the present comparison, it is worth briefly explaining why journals operating under a Subscribe-to-Open (S2O) model are not included in our dataset. While S2O enables content to become openly accessible without direct author-facing publication fees, the continuation of OA remains contingent upon the willingness of subscribers to renew their subscriptions. As a result, OA is not guaranteed as a permanent feature of the journal and may be reversed if sufficient subscription revenue is not maintained. In contrast, our definition of Diamond OA requires that OA is provided independently of subscription income. For this reason, S2O journals are not considered Diamond OA within the scope of the Belgian Diamond OA Landscape dataset.[20]
The reverse perspective is equally informative. Looking at the journals that are unique to the Belgian Diamond OA Landscape dataset reveals a number of patterns that might help explain the differences between both datasets. A first observation is that a substantial number of journals hosted on institutional and community-based publishing platforms such as OpenEdition Journals, Openjournals.nl, Open Journals UGent, OJS UCLouvain, and PoPuPS (ULiège) were absent from the ALMASI dataset. This suggests that, as expected, the bibliometric approach is less effective at identifying journals published through smaller institutional and community-based Diamond OA infrastructures, and is more suited for identifying journals and publishers that are well represented in international metadata sources.
| Platform | Shared by both datasets | Only in Belgian Landscape dataset |
|---|---|---|
| Openjournals.nl | 0 | 4 |
| OJS UCLouvain | 5 | 11 |
| PoPuPS (ULiège) | 5 | 18 |
| OpenEdition Journals | 10 | 10 |
| Open Journals UGent | 10 | 12 |
Tabel 1. Shared and unique journals by publishing platform.
A further explanation for the observed differences between the two datasets can be found in disciplinary orientation.[21] Only 36% of HSS journals are found in both datasets, compared with 46% of SET journals and 53% of BMS journals.[22] This pattern partly reflects the strong presence of HSS journals on institutional and community-based publishing platforms, which themselves are underrepresented in the ALMASI dataset as shown above. This observation is also consistent with broader scholarly communication patterns, as SET and BMS journals tend to be better represented in large international indexing and metadata databases, making them easier to identify through bibliometric approaches.
| Disciplines | Shared by both datasets | Only in Belgian Landscape dataset | Total |
|---|---|---|---|
| HSS | 46 (36%) | 83 (64%) | 129 |
| SET | 12 (46%) | 14 (54%) | 26 |
| BMS | 8 (53%) | 7 (47%) | 15 |
Tabel 2. Shared and unique journals by discipline.[23]
Publication language, on the other hand, does not appear to provide a convincing explanation for the differences between the two datasets. To classify the journals by publication language, we applied the same broad categories as used in the ALMASI scoping report:[24]
- English: journals publishing exclusively in English;
- Multilingual: journals publishing in English and at least one additional language;
- Other languages: journals publishing exclusively in non-English languages.
The proportion of journals identified by both datasets is remarkably similar across English-language, multilingual, and other-language journals, suggesting that the bibliometric approach does not systematically favor or overlook journals based on publication language.
| Publication languages | Shared by both datasets | Only in Belgian Landscape dataset | Total |
|---|---|---|---|
| English | 17 (37%) | 29 (63%) | 46 |
| Multilingual | 34 (39,5%) | 52 (60,5%) | 86 |
| Other languages | 12 (40%) | 18 (60%) | 30 |
Tabel 3. Shared and unique journals by publication language.
Comparing the Belgian Diamond OA Landscape dataset with the ALMASI dataset provided valuable insights into how methodological choices, including data sources, inclusion criteria, and validation procedures, shape the representation of national publishing landscapes. While large-scale datasets make cross-national analyses possible, national studies can provide contextual knowledge and journal-level verification that are difficult to achieve at scale. The comparison revealed discrepancies related to journal activity status, national affiliation, and, most importantly, the classification of OA models. It also showed that journals published through institutional and community-based publishing infrastructures, as well as Humanities and Social Sciences journals, were less consistently represented across both datasets. The Belgian case therefore highlights the need for further reflection on how Diamond OA journals are identified and classified, particularly when the aim is to capture diverse and locally embedded publishing initiatives.
[1] Laura Mesotten and Demmy Verbeke, The Belgian Diamond Open Access Landscape, June 23, 2026, https://doi.org/10.5281/zenodo.20760254.
[2] Version 2 of the dataset was published on October 9, 2026 and is available via https://doi.org/10.5281/zenodo.20760253.
[3] Zehra Taskin and Mikael Laakso, ALMASI Dataset of Diamond Open Access Journals, December 11, 2025, https://doi.org/10.5281/zenodo.17897299.
[4] ROAD (Directory of Open Access Scholarly Resources) is an ISSN-based international database maintained by the ISSN International Centre that provides metadata on Open Access scholarly resources. See: https://www.issn.org/services/online-services/road-the-directory-of-open-access-scholarly-resources/.
[5] OpenAlex is an open bibliographic database containing metadata on scientific publications, authors, institutions, and funders.
[6] OpenAPC is an open initiative that aggregates data on Article Processing Charges (APCs), Book Processing Charges (BPCs) and data on articles published under transformative agreements.
[7] DOAJ (Directory of Open Access Journals) is a community-curated index of peer-reviewed Open Access journals. Although DOAJ metadata were used during the construction and validation of the ALMASI dataset, DOAJ inclusion alone does not appear to determine whether a journal is captured by the ALMASI methodology. In the Belgian Diamond OA Landscape dataset, 18 of the 49 journals (37%) indexed in DOAJ are absent from the ALMASI dataset.
[8] Mikael Laakso and Zehra Taskin, D1.1 Scoping Report on Non-for-Profit Publishing Ecosystems, December 29, 2025, https://doi.org/10.5281/zenodo.18257718.
[9] DDH (Directory of Diamond Journals) is a community-maintained directory of Diamond Open Access journals.
[10] OpenEdition Journals is a publishing platform and infrastructure for Open Access journals in the humanities and social sciences, operated by OpenEdition.
[11] All consulted data sources are described in the dataset’s About section.
[12] Journals that explicitly state that publication has ceased, that have not published any content within the previous five years, or that have been launched but have not yet published any content are considered inactive and are excluded from the dataset.
[13] These inclusion criteria are described in more detail in the dataset’s About section.
[14] All tables and figures presented in this article are based on the combined set of Belgian Diamond OA and Belgian delayed OA journals.
[15] Laakso and Taskin, D1.1 Scoping Report on Non-for-Profit Publishing Ecosystems, 132.
[16] Simone Angioni et al., “Charting the Landscape of Italian Diamond Open Access Publishing,” Quantitative Science Studies 7 (June 2026): 564–84, https://doi.org/10.1162/QSS.a.466.
[17] The duplicate record concerns Revue Quetelet / Quetelet Journal, which is listed in the ALMASI dataset under two separate records: Revue Quetelet + Quetelet Journal (ISSN 2593-9149, eISSN 2593-9157) and Revue Quetelet / Quetelet Journal (ISSN 2034-9378, eISSN 2034-9386). The latter record is also registered in the ISSN Portal, albeit under the title Chaire Quetelet. It presumably refers to an earlier registration of the same journal and was therefore treated as a duplicate.
[18] Some journals met more than one exclusion criterion and were therefore counted in more than one category. In particular, the category No Belgian affiliation overlaps with Gold Open Access with APC (n = 3) and Paywalled access (n = 1). As a result, the counts for the individual exclusion criteria should not be added together, as this would exceed the total number of excluded journals (n = 92).
[19] Recent recommendations of the OpenAlex Colors Working Group highlight important limitations of current Open Access classification systems. In particular, the report argues that widely used OA classifications often conflate access conditions with publishing and funding models, making it difficult to consistently represent more complex OA arrangements and to distinguish between different publication models on the basis of bibliographic metadata alone. See Juan Pablo Alperin et al., Recommendations for Revising the Open Access Classification System in OpenAlex, August 14, 2026, https://doi.org/10.5281/zenodo.21934115.
[20] Some studies and mapping exercises adopt a broader definition of Diamond OA and do include S2O journals, e.g. Niels Taubert et al., “Mapping the German Diamond Open Access Journal Landscape,” Minerva 62, no. 2 (2024): 193–227, https://doi.org/10.1007/s11024-023-09519-7.
[21] The disciplinary categories were assigned by the authors as part of the Belgian Diamond OA Landscape study, based on the primary focus of each journal. As a result, some classification decisions may be open to interpretation.
[22] The totals by discipline exceed the number of journals because some journals were assigned to more than one disciplinary category and are therefore counted more than once.
[23] For Tables 2 and 3, percentages are calculated relative to the total number of journals in the respective category.
[24] Laakso and Taskin, D1.1 Scoping Report on Non-for-Profit Publishing Ecosystems, 135.