We all agree on the benefits of using PIDs, persistent identifiers, the use of standard metadata such as DOIs for objects, ORCIDs for people, RORs for institutions) to reduce ambiguity and to maximise interoperability. But the problem with metadata is always how to make sure it is used, and used correctly. Simply agreeing on the solution isn’t enough: somebody, or something, has to add the right tags.
The More Brains Collective produced an interesting paper for Ireland’s National Open Research Forum, NORF, on how to implement a national strategy for PIDs. Their argument looks sound, but the proposed execution has, I think, a few holes.
More Brains produced a cost-benefit analysis for using PIDs across all 25 or so institutions in Ireland. Their methodology is simple. Estimate how long it currently takes researchers to enter data such as grant numbers and author ORCID IDs; estimate how much time could be saved by only entering the metadata once; and hey presto, they estimate you can save €1.6m over five years. Sounds like magic.
How do they work it out? Pretty simple. They estimate it takes researchers 4000 days of staff time each year to enter just four items of metadata, since it is often added more than once. This costs €1.8m. Instead, they propose a “central support service” of three people, costing around €0.35m per year, plus costs of around €200K per year for implementation.
The support service would comprise a coordinator, a technical community manager, and an engagement and outreach manager – three chiefs, but no Indians. From the job titles, it would appear that these three people explain and evangelise about the wonderful world of machine-readable metadata. But I don’t think any allowance is made for anyone other than the academics themselves to enter the data.
Will it work? I have my doubts. Academics are notorious for not following rules. They complain about bureaucracy, and this will be just one more item. To get academics to enter any metadata is challenging, and PIDs will be, they will claim, more complex.
What happens in practice, today? One US institution described how they email all their researchers whenever they publish a new article. The email invites the research to deposit the article in the institutional repository. However, their response rate to these emails is around 25%: three quarters of the emails are ignored. The University of Oxford tries to make it as simple as possible for their researchers to add articles to the repository; they have refined the process so that the researcher only has to add the minimum metadata, such as author and title, and the repository team then does the rest, identifying the article and filling in the metadata gaps. At this stage, the enterprising Oxford team have written scripts that find relevant metadata from elsewhere, and provide the other necessary parts of the jigsaw.
Of course, the researcher is the beneficiary of all this activity, because their article becomes more discoverable, and this leads to higher citations. Yet all the repository managers I have spoken to are united on one point: that researchers are the worst people to compile metadata.
Here is an alternative, and I think a more practical suggestion. Since every institution already has a repository, with staff who understand metadata and how important it is, what if the repository was tasked with filling in the gaps for the researchers and hence creating the machine-readable universe we all benefit from? This largely manual operation is, as we speak, in the process of being automated by tools such as those implemented by Oxford.
Instead, what if repository staff (who already exist in each institution) ensured that the relevant PIDs were input, rather than the researchers? Some work would be required to adjust the workflow for grant proposals, and so on, but the repository would follow central instructions, and would be far more likely to complete the metadata requirements accurately.
Certainly, there would be staffing implications in the short term, but I would expect this to reduce as the quality of metadata improves. This would provide a key role for repositories who could claim with justification that they are moving the institution towards Open Science and FAIR compliance. Trusting the academics to complete all the relevant fields would, I think, be unpopular and would not achieve its aim.
Alongside the manual tagging, I would expect some institutions and organisations to provide software tools that continue to automate the adding and checking of metadata, so the involvement of humans rapidly becomes one of checking and approving rather than manual copying and pasting.
Effective metadata strategies require an agreement on the principles, but also a feasible method for getting the metadata added. This strategy looks to me to be a potential way of using existing resources. I’d welcome any feedback!
Leave a Reply