November 1, 2004

Harvesting the benefits of Open Access

Governments around the world are recommending the establishment of electronic archives to provide Open Access to national scholarly output. But it remains unclear how best to create and manage these archives and how to ensure that they will be effectively used by researchers. A former publisher who's now an academic, Fytton Rowland recently led a study that recommended a 'harvesting model' for Open Access provision in the UK. He talked to Open Access Now about different models of Open Access archiving.

Fytton Rowland is an unusual example of a publisher turned academic. With over two decades of experience in scientific publishing, he has played the publishing game from both sides of the fence. "I spent the first half of my career working for several different learned society publishers," explains Rowland. He worked both in traditional scholarly journal publishing and database publishing. "Most of that half of my career was with the Royal Society of Chemistry. I was publications production manager and involved in their first experiments with electronic publishing." Over a decade ago Rowland crossed the divide and joined the staff of Loughborough University in Leicestershire, UK, initially as a research fellow and then as a lecturer. He is responsible for running the undergraduate BA Honours degree course in publishing, which covers both traditional and electronic publishing.

Rowland is also the founder of the Electronic Publishing Innovation Centre (EPIC), a research group of collaborative academics and librarians. "The virtual centre was set up jointly between the Department of Information Science at Loughborough University where I work and the Information and Library Services at Cranfield University, where my wife, Hazel Woodward, is the librarian," he explains. "We were both interested in electronic publishing issues and scholarly publishing issues. We established a collaboration between our institutions because both had a track record of doing research in the scholarly communication field. Unlike the other research groups in the department our group incorporates staff from both universities." EPIC carries out research into a range of issues related to electronic publishing. "We have quite a lot of people at Loughborough and Cranfield who are interested in different areas of electronic publishing," notes Rowland. "So for any given project we try to choose the right people with the right expertise - be it legal issues, information technology or metadata analysis."

"The aim was to advise JISC on how it might proceed in providing a national system of Open Access repositories, to take in both material held in institutional repositories and material that is published in Open Access journals, like the ones published by BioMed Central for example," says Rowland. "We wanted to look at different models and evaluate them from various points of view - technical, legal and metadata areas, and so forth - and then make recommendations about what might be the best model." The Centre has just published its 121 page report, entitled "Delivery, Management and Access Model for E-prints and Open Access Journals within Further and Higher Education" which proposes a number of recommendations to JISC about favoured access models.

The report concerns both e-prints, digital duplications of academic research articles that are made available online to permit increased access, and articles published in Open Access journals. The report avoids commenting on which strategy ('self-archiving' or Open Access journals) is the best to achieve increased access and Rowland refuses to be drawn into the debate. "Becoming too closely wedded to one particular path is unwise," he says. "I think we have to pursue both roads at the same time, supporting Open Access journals and encouraging people to mount material on Open Access servers."

The EPIC study identified three models for Open Access provision in the UK. First is the 'centralised model', whereby e-prints of articles are first deposited directly into a national archive and then made accessible to users and service providers. Second is the 'distributed model', whereby e-prints are deposited in any one of a distributed network of institutional, subject-based and Open Access journal archives, metadata from which can then be harvested and made accessible to users by service providers. And third is the 'harvesting model', a variant of the distributed model in which the harvested metadata are first improved, standardized or enhanced before being made accessible to users and service providers.

The report concludes that the centralized model should not be adopted in the UK, as it is costly and would omit existing archived material. The report refers to the centralized model as the "wrong way round", because academic and institutional culture would make it hard to fill a central archive effectively. "What we recommended was that a harvesting model would meet the requirements best," says Rowland. "The full-text data remains on institutional servers, or servers belonging to a particular subject discipline or Open Access journal - wherever they might have started off. The central server would provide a harvester that would go round and look regularly at all of the new information on the distributed servers, bringing it to the attention of the staff at the centre who would hold the metadata and would, if necessary, amplify, standardize or augment the metadata." Rowland explains that this model has the advantages of both centralized and distributed models. It does require that all archives be interoperable, an aim that is ensured by the Open Archives Initiative-Protocol for Metadata Harvesting (OAI-PMH), a standardized technical format that allows servers to extract information from distant sources. The report's farmed harvesting model allows a mechanism for enhancing metadata beyond the basic standards of OAI-compliancy if more sophisticated information is required in some fields.

Rowland points out that the EPIC study was being conducted at the same time as the House of Commons committee inquiry. Although the academics were not privy to what the politicians were going to say, they came to similar conclusions. "We came up with the conclusion that a distributed system, rather than a centralized model, was probably more practicable. But we added this additional sophistication of having a central harvesting of the metadata."

"I think that everyone recognizes that the key issue for institutional repositories is how to populate them," notes Rowland. "From the technical point of view it is quite easy to set up an institutional repository; the hardware is not unduly expensive and open source software like Dspace or E-prints is readily available. But everybody who has looked into this issue agrees that the key problem is how to persuade the academic authors to actually put their stuff in. This is one of the human issues that our report analyzes. In order to make sure that lots of academics put lots of material into these servers you have got to make it very easy to do so. Any form of legalistic or technical barriers or obstacles will simply mean that people don't bother. Academics have busy lives these days, we have increasing numbers of students to teach and we are under increasing pressure to do research. If it is not incredibly easy to do we just won't do it because it won't be a high priority."

"It's a question of carrots or sticks," comments Rowland. Both the House of Commons committee and the NIH feel that the way to provide free access to publicly funded research is to mandate deposition in Open Access institutional or central archives. "NIH is supporting legislation through the US Houses of Congress to similarly mandate that NIH-funded research is deposited in PubMed Central, and the UK committee is advocating that the research councils should mandate Open Access archiving. Effectively what they are advocating is that there should be sticks," he explains. He notes that a number of funding agencies around the world (including the Wellcome Trust, the Max Planck Institute and the Howard Hughes Medical Institute) have expressed views sympathetic to Open Access archiving and are moving in that direction.

"But another way is to offer carrots as well as sticks," suggests Rowland. He proposes that institutions might provide staff to assist academics with the task of putting material into archives. "Most commonly these staff are based in university libraries, though there is no reason why they might not be in the university computing service instead. If you look at the costs for institutional servers, by far the biggest chunk is staff costs, and when you ask what these people are doing the answer is that these are staff whose job it is to pro-actively go and find material around the university that needs to go into the server." He feels that the harvesting model will allow universities to develop incentives and mechanisms to fill their local repositories. "In due course most UK universities will have institutional repositories which will be part of the basic infrastructure of the university and will be covered, like the library or computing services are, from university central funds. I think that is a realistic hope."

Rowland insists that the procedure for inputting material must be as easy as possible. "The servers will have to be as tolerant as possible about the different formats that might go in and should not expect the individual academic members of staff to do it all themselves. The university must provide some form of central provision within the campus that will basically take whatever the academic throws at them and go through the technical work of mounting the material into the OAI-compliant server." The EPIC report implies that JISC would then play the role of central harvester and content provider, with the British Library offering a 'back-up' repository for authors who were independent of universities or research institutions. "It's not for us to say which organization should be the national service provider, but it was envisaged that a national service provider would network with similar service providers in other countries," comments Rowland.

The EPIC report has lots to say about how JISC might coordinate the national archiving exercise and encourage academics to participate. JISC is likely to accept many of the study's recommendations, so British universities can expect to contribute to the creation of a national 'harvesting model' for Open Access in the years ahead.