Reference Reviews

Péter's Digital Reference Shelf

December 2009

Title: DOAJ — Directory of Open Access Journals
Publisher: Lund University Libraries
URL: http://doaj.org/
Tested: December 1-3, 2009
Cost: free

The DOAJ database, with bibliographic information about nearly 5,000 open-access journals, offers more than its name implies, by virtue of also having searchable traditional bibliographic data, keywords and abstracts for 331,000 articles of 1,725 open-access scholarly journals. The software offers good browsing options for the journal records, but the search and output features should be enhanced. Adding records about journals with delayed open access (of 6-12 months moratorium) would significantly enhance this excellent database.

THE CONTEXT

Open-access articles, conference papers, books and journals are increasingly important in light of the ever-increasing, and often unrealistic purchase prices, subscription fees and service costs of interlibrary loan and document delivery services of scholarly source documents.

Strangely, with the exception of Ulrich’s Periodical Directory, none of the subscription-based serials directories offer an option to search for open-access journals. In the open access arena, neither does the otherwise very good and free Journalseek database. It does yield 273 hits when searching in all the record fields for the term "open access", but it is not a field-specific search (so there may be false hits), and a very low number from 95,320 titles. There is an excellent free resource dedicated to open-access journals, Open J-Gate from Informatics India Ltd. It covers about 20% more journals than the Directory of Open Access Journals (DOAJ), but it is not in the same league as DOAJ. Its primary purpose is to find open-access articles in 6,000+ journals, many of which are not open access, but simply offer some open access articles. In addition, it does not have the typical metadata set of journal directories (ISSN, country, language and frequency of publications, their start year, CODEN, coverage by indexing/abstracting databases). In all fairness, many end-users can happily live without the desire of knowing, let alone searching by, most of these attributes, and DOAJ does not have all these metadata elements either.

There are two important other directories that focus on a different genre, but should not be ignored in the context: ROAR, the Registry of Open Access Repositories from the University of Southampton, which pioneered the idea, and OpenDOAR, the Directory of Open Access Repositories from the University of Nottingham, and the University of Lund (the developer and host of DOAJ). Both are open access, and have information about 1,500 scholarly repositories.

It is also very appropriate to mention the comprehensive high quality Open Access Bibliography - Liberating Scholarly Literature with E-Prints and Open Access Journals of Charles W. Bailey, whose various systematically updated open-access bibliography series are a gift for anyone interested in open-access works, and represent a pride of our profession.

THE CONTENT

There are many content features that deserve acknowledgement, such as no limitations by language and/or country of publication, which is important for example for Brazil, a front country in the open source and open-access movements, where many of the worthy publications are in Portuguese. There are very few constraints that I disagree with, and discuss in this review.

Scope

DOAJ defines its scope with exceptional clarity, explaining what qualifies as open access journal in terms of accessibility, content, subject area, frequency of publication and other criteria, such as having an ISSN, which is kind of a commitment (on behalf of the publisher of the journal who has to apply for it), and a filter for qualifying the publication as a serial (by virtue of being accepted by ISDS, the International Serials Data System) even in the Web-sphere where anyone can start publishing a "scientific journal".

Very rationally, it does not restrict itself to peer-reviewed journals, but also includes ones that have "only" editorial control, which –in my opinion– makes many of them of better quality than some peer-reviewed journals in the disciplinary area. I know that this is anathema in academia, but in my publishing career I have received much more useful criticism, corrections and precise recommendations from my editors than from peer reviewers.

I do see both sides of the coin because I do about 25 reviews per year for 4–5 peer–reviewed academic journals and conference proceedings, and try to be constructive and informative for the authors and the editors.

One of the restrictions of DOAJ is that the publication is expected to be published generally more frequently than once a year. This could disqualify some important and open-access proceedings of conferences that are held once a year or every second year. This is the case with the Proceedings of the Hawaii International Conference on System Sciences (HICSS), which are available for the past 15 years through the Web site of the IEEE Computer Society for free. (This may seem to be a geographically myopic example, but it is not biased as I have not been involved in any way in HICSS, which has been a top-ranked conference by various rating systems), still it is not in DOAJ. True, the letter J stands for journals, but the scope refers to periodicals, and indeed there are a good dozen proceedings, not necessarily conference proceedings covered by DOAJ.

A more neutral but fictive example could be that this policy would exclude from coverage a possible series of Annual Review of Open Access Scholarly Systems and Services. The annual review series are among the top listed serials in many disciplines (and although none of them are open access - yet , they have very reasonable prices for the value they deliver).

The other restriction is that the journal must be open access in a sense that all the papers become open access at the time of publication. It is a noble idea, and I understand on the one hand the puritanistic approach but many users would be very satisfied being led to journals that do have a 6-month or 1-year moratorium — which may have already expired when the item is looked up by the searcher. This policy excludes hundreds of high quality scientific journals that have a rationally delayed open access policy, and a huge collection of widely read and widely cited papers.

This strictness of DOAJ may leave unaware users of millions of free articles, such as those 2 million open-access papers that are available through HighWire Press alone - directly or indirectly. It is worth illustrating this disagreement to drive home my point, indicating the number of open-access articles and their percentage of the total articles for some of the most important delayed open-access journals with the largest collection. My data refer to the early November, 2009 status.

I don’t need to fill space in my column, and I could have attached these data in a spreadsheet, but I intentionally spell them out in the text in case the reader does not click for the attachment.

The top periodicals with 6-month moratoriums and 96-99% of the total number of articles include the Proceedings of the National Academy of Sciences (98,381), Annals of Internal Medicine (49,601), Journal of Bacteriology (46,649), Journal of Virology (34,159), Applied and Environmental Microbiology (29,450), Infection and Immunity (27,724), Journal of Clinical Microbiology (23,209), Journal of Neuroscience (22,343), Journal of Cell Biology (21,991), Journal of Experimental Medicine (21,555), Antimicrobial Agents and Chemotherapy (19,473), Molecular and Cellular Biology (19,372), Genetics (16,514), Journal of Cell Science (15,800).

The top journals with 12-month moratoriums and 95-97% of the total number of articles include Cancer Research (46,776), Chest (35,864), Blood (34,955), Circulation (32,134), Journal of Physiology (29,294), Plant Physiology (26,115), Journal of Dairy Science (23,678), Radiology (23,671), American Journal of Roentgenology (23,848), Journal of Animal Science (22,531), Clinical Chemistry (21,416), Journal of Nutrition (20,351), American Journal of Clinical Nutrition (17,716), Microbiology (17,221), Annals of Thoracic Surgery (16,618), Journal of Dental Research (13,721), Circulation Research (13,621), Journal of General Virology (13,243).

It is not the number of these journals that is important to be added to DOAJ but the number of open access articles (which are around 1 million for thosie I listed), and the influence of these journals. A good compromise would be to include in DOAJ the journals which have delayed open access policy, and allow the users to exclude/include those. Considering the clout, the impact factor of these journals, which make them top ranking in their fields, could be another argument to include these journals in DOAJ — but the lawyer in me suggest to just say: I rest my case.

Database size

DOAJ has records for 4,475 open-access journals, and by the end of 2009 it will certainly have information about more than 4,500 titles. This is not just obvious, idle talk — the overall growth rate from 2002 to the end of 2009 is very impressive as indicated on my simple improvised chart. In addition, the customizable counter shows that in the past 30 days entries were created for 70 new journals.

The comparison by the number of journal titles with alternative sources is not as easy as it may seem, for several reasons. In the subscription-based Ulrich’s Periodicals Directory, there are information about 4,401 open access journals, but this is an underestimate because Ulrich’s follows the principle of latest entry cataloging (as opposed to successive entry cataloging). In the former, all the journal title variations are listed in a single record, in the one created for the most current title.

DOAJ and the other directories do not treat and count the same way the journals that changed titles, and the consistency of inclusion is not perfect either. In DOAJ there are two entries – correctly- for two journals of the medical Library Association. One for the Bulletin of MLA, and the other for the Journal of MLA, the successor title of the former. In the digital era the entire run of both periodicals became open access from Volume 1, Number 1.

In Ulrich’s there is an entry only under the current title, which does include information about the former, but it is not counted as a separate open access journal. For example, Acta Orthopaedica Scandinavica was acquired by Taylor and Francis, and the name changed to Acta Orthopaedica. There is only one entry for this journal in Ulrich’s, but two entries in DOAJ. Unfortunately, it is a duplicate entry – except for the start year. It seems to be due to an error at the new publisher’s site. It is not possible to determine how many other duplicates there may be from this publisher in DOAJ, but such errors adds to the difficulty to compare the size of directories in terms of journal titles. One record less here, and another one there, may not seem to be a big deal in counting the total number of journals covered by different directories, but actually, it is.

In Open J-Gate the number of open-access journals is reported to be more than 6,100 but many of the journals are only partially, very partially open access, such as Searcher magazine which indeed has a few open access articles in every issue, but it is certainly not an open access journal. It is great to find open-access articles from 6,100 journals through Open J-Gate, but the size comparison in terms of open-access journals with DOAJ would not be fair.

I have spent several years dealing with automating serials control procedures and the practice of title changes seems to be a favorite one among many publishers, including ALA which should have known better before changing the title of what we know now as School Library Media Research at least four times. This makes it difficult to estimate how much larger Ulrich’s is than DOAJ in terms of open-access journals, but it is certainly not smaller as the reported number of OA journals would suggest.

Among others, I tested 13 journals published by the European Geoscience Union and found correct matches for 12 journals. The exception was the journal Solid Earth. I also tested the availability of 20 open-access journals in Library and Information Science, and there were records for 19 of them – a very good ratio. The only title missing from DOAJ was the Chinese Librarianship, but when verifying my search later it did appear, producing a 20/20 hit rate. Overall, the open access journals published in China and available in DOAJ (14), seem to represent a fraction of the ones that would probably qualify for inclusion. Ulrich’s has information about almost twice as many open-access journals published in China, but even that may be a number significantly lower than the actual one. For my samples for the immediately open-access journals (i.e. excluding books) hosted by HighWire Press, the availability ratio in DOAJ was only 60%.

Subject coverage

DOAJ has excellent broad subject area coverage. Journals are classified under 17 broad subject categories. This is a reasonable categorization, except perhaps to have a separate main category for Arts and Architecture (111 journals), but not a separate main category for Humanities, where Languages and Literature alone has 160 journals and History has 130.

This oddity is accentuated by the fact that there is a much larger number of open-access journals in all the Humanities fields combined than in the Arts and Architecture category. It would be good to show the number of journals for the 17 main categories, as it is done for the subcategories, where the only distraction is that under the Social Sciences main category there is a Social Sciences subcategory with 194 journals – and this needs to be corrected. One journal may be assigned to several categories, such as Psychology and Sociology. On the average, 1.2 main categories are assigned to journals, so this subcategory problem within the Social Sciences could be easily corrected.

I created a small table to indicate the share and total number of journals across the 17 main categories to see the distribution, but these will obviously change. Nevertheless, it is quite informative.

Not surprisingly, Health Science journals have the largest share (23.46%), followed by the Social Science main category (18.43%). Biology and Life Sciences (8.68%) and Technology and Engineering (8.33% are quite close). The low number and share of open access journals in Physics and Astronomy (2%) , and in Chemistry (1.96%) may seem surprising, especially that of the former category as astronomy and physics have had the strongest open access culture through decades of distributing and self-archiving the pre-print and re-print versions by the physicists and astronomers. However, on second thought, these two fields needed the least open-access journals, because their researchers (especially those in astronomy) made available far the most intensively and far the longest period of time the open access preprints of their papers - published in subscription-based journals.

The best way to get a panoramic view of breadth and depth of the subject coverage is to expand the categories, and scan across the horizon. DOAJ makes this very swift and simple, and the panorama is very impressive and sometimes surprising for some specific disciplines.

Geographic coverage

There are open-access journals in DOAJ produced/published in more than 100 countries. This is quite good for internationalism. The U.S. has the largest share with 965 (21.65%) open-access journals. However, the combined European presence is almost twice as large. Brazil has the second rank with 405 (9.05%) open-access scholarly journals.

The reason that it is not surprising to me is my modest familiarity with the success of the outstanding SCIELO (Scientific Electronic Library Online) open-access service that was developed in Brazil and hosted by Brazil not only for Brazilian journals but for many other journals published in South American, Central American, Caribbean, and Iberian countries primarily for their community of scholars. The efforts that Brazilian information technology institutions, especially BIREME, and specialists have put in have enhanced the most popular, highly sophisticated information retrieval software, CDS/ISIS, originally developed and distributed by UNESCO since the 1980s. It is also a testament to the competence in and enthusiasm for the open source and open-access movements in the country, not to mention the open-source Javascript version of the software ABCD that was just released in its first production version as I was finishing this column. My e-mail was buzzing with news about this somewhat delayed, but very important event.

As for the distribution of the other 3,100+ open access journals by country of publication in DOAJ, there is a not very prominently positioned but informative table showing how other countries have been progressing with launching open-access scholarly journals in the past eight years. Not surprisingly, the top 20% of the countries provide 79.05% of the open-access journals covered by DOAJ. I wish there were a similar table by the distribution of the journals by language which is not directly searchable. It is possible to search for, say, Hungarian, language open access journals, but because the search is not field specific, Acta Zoologica Academiae Scientiarum Hungaricae comes up as a match because the publisher is the Hungarian Academy of Science, but the language of the journal is English only.

THE SOFTWARE

The DOAJ site is well-designed and simple to navigate, although the section for authors is somewhat redundant and confusing. However, it offers the big advantage of informing potential authors about the fee –if any- of making an article open access. Then again, if not all the research articles are immediately open access at publication time, the journal is not supposed to be covered. Tracing the link for further information about the fee did not provide any clarification in the few cases that I traced (not if I were interested in paying for getting published).

The journals can be browsed by title and subject, which is a good idea, as in my opinion most good searching starts with browsing to learn about the bibliographic conventions used in creating the data.

The good browsing options also compensate to some extent for the disappointing limitations of the search module. There is no way to truncate a search to the stem of the words, such as geo* for geology, geologic, geosciences, geophysics, or geophysicae. It adds insult to injury (or at least frustration) that it takes several minutes to get the not found message when trying to truncate.

Similarly inconvenient is the restriction that only exact phrase searching is possible if more than one term is used. Most information systems have switched to interpreting the space in a query to be a Boolean AND, which is a much more liberal approach, especially considering the difficulty of knowing the exact title of journals, and guessing how it appears in the database. When searching by journals, no field can be specified, i.e. a journal title search cannot be conducted, and if the title of the journal is not distinct enough, and coincides with descriptors assigned to it, those will be also retrieved. Of course, one can look at it from the perspective that the primary search criteria are topical (microbiology, diabetes) not known items, and it is useful to bring up all the open access journals whose keywords and/pr publisher include the search term. There is no option to select records from the result list of a search, and the output cannot be saved in one of the standard bibliographic export formats, such as EndNote, RIS or MARC Communications format, for making it easier to import those into a bibliography, or into a catalog that would make users aware of the existence of many of the high quality open access journals.

DOAJ compensates to some extent for these software limitations by making possible to search the entire bibliographic records and abstracts at the article level for 331,000 papers published in 1,725 journals. Not all of them have abstract available, but in my test, more than 260,000, nearly 80% had, and they significantly improve the discovery rate of potentially relevant articles, as well as the selection process from the result list through compact and informative summaries.

This is a great addition to a directory, and the introduction of the Find article option deserves kudos for allowing field-specific searches by article title, journal title, ISSN, author, keyword and abstract – and the use of the three basic Boolean operators for one third of the journals. Most importantly, all the records lead the users to the full-text articles hosted at the site of the publishers or their digital facilitators.

It is no wonder that on the days I was testing the system for this review, and watching the number of visitors counter, the daily rate was above 6,000, and the number of active users was never below 45 when I was online with DOAJ. It was an excellent idea to launch this project, and it very well-deserved the start-up financial support it got from the Open Society Institute of the Soros Foundation, and SPARC Europe, the alliance of European Research libraries, library organizations and research institutions. The same is true for the continual support from Lund University, the National Library of Sweden, the Swedish Library Association and INASP, the International Network for the Availability of Scientific Publications.

I hope that the DOAJ developer team will consider enhancing the database by including scholarly journals with delayed open access. DOAJ is one of the few universally relevant, tangible and precious services that Steven Harnad, Peter Suber, Charles Bailey, Gary Price and his fellow editors have been advocating for and advised us about with relentless passion and competence for more than a decade. The DOAJ system received the SPARC Europe Award for Outstanding Achievements in Scholarly Communications earlier this year, and it well deserved the recognition.