The increase of Open Access (OA), as a percentage of all peer-reviewed research articles, is not in dispute. There are even scientific studies on the rate of increase and its significance. Given the ascendency of OA, researchers and libraries must address some basic issues resulting from a movement that has no single, unified standard. One major concern involves the Green OA model, which consists of self-archiving work into various open access repositories, like Cornell’s arXiv. These repositories have been developed independently, using a wide variety of IT systems and protocols. While many data providers are OAI-PMH compliant, the adherence to this protocol at the individual record level is inconsistent.

Language & Platform Problems

On a basic user level, there are barriers to e-document reading that were never at issue for printed journals. For example, arXiv accepts digital documents in HTML, PDF, PostScript, and TeX. While it’s true that some of these page description languages (e.g., TeX) are superior for certain scientific publishing tasks (e.g., math formulae), not every researcher or librarian has the means of opening or reading every format. The source of compatibility problems sits at the root of open source development in general. When Open Access arose, it was in response to the digitization of content—and the rising costs of traditional publishing. However, there was no one solution, and various implementations were developed independently. Differences in operating systems, database environments, content management, and security infrastructure each add to the degree of difficulty in making repositories interoperable—and globally transparent to researchers and librarians.

Mapping the Metadata Muddle

All this pales in comparison with the issue of metadata. Being able easily to find a particular paper—and then find other relevant research, regardless of what repository it’s in—is a noble goal, beset by the idiosyncrasies of the isolated IT approach to Green OA. If a different label or tag is used for the same type of data, finding all the data is problematic. That appears to be changing. The SHARE initiative has taken on the challenge of establishing a common set of descriptors, and a common data structure to support them. Led by the Association of Research Libraries (ARL), the Center for Open Science, and others, SHARE’s mission is to create “a free, open, data set about research and scholarly activities across their life cycle.” We spoke with ARL program director Judy Ruttenberg about SHARE and its prospects. “I’m very optimistic about it,” she said. “We have more than 100 providers with whom we are harvesting metadata. At our upcoming Community Meeting, we’ll be launching a new program on shared curation that will involve librarians and data professionals working directly to enhance this metadata. Of course at scale we will need to automate as much as we can with machine learning and statistical data cleaning.” Ruttenberg noted that institutions have a particularly strong motive for backing this effort. “The reason people put material into an open repository is to have that material discovered,” she said. “If we can aggregate all their metadata into an open data set, and give them the means of linking to related data, then they will have every reason to participate.” The work will require full participation. “SHARE is a very ambitious endeavor to solve a big challenge—the lack of accessibility, openness, and interoperability of data about scholarship,” Ruttenberg said. “While I'm very optimistic about the path we’re on, SHARE is the kind of challenge we need the entire community to embrace and invest in—in order for it to succeed.”

Global Initiatives

We also spoke with Kathleen Shearer, the executive director of the Confederation of Open Access Repositories (COAR). Originally begun in 2009 as an EU-sponsored project, COAR has become an international voice for over 100 libraries, universities, research institutions, government funders, and others in the repository community. COAR’s objectives include promoting strategies to fill repositories with content as well as supporting the adoption of next generation functionalities and technologies for repositories. The group also works with the repository community towards standardization and interoperability. “Rather than try to engage with every individual repository—and there are thousands of them—we engage primarily with the networks, which are the ‘harvesters’ of OA content,” Shearer said. “These include entities like OpenAIRE in Europe and similar groups in the U.K., Latin America, SHARE in the US/Canada, and elsewhere.” One of COAR’s goals is to encourage mutual ‘data exchange’ agreements between these networks as a means of reaching a common metadata approach. Research libraries and their users can only stand to gain from these efforts. “Institutional repositories will no longer be standalone entities for just their own work,” Shearer said. “They will be contributing to (and benefitting from) a global collection. The users will benefit of course. They will have better discoverability and ability to see the relationship of content types and data that were not readily available before.”