Who knows about repositories? They are a fundamental part of the scholarly world: pretty much every institution of higher education around the world has a repository. Yet what they are, how they are run, and why they exist are often misunderstood or simply not known. You may be surprised by what you read.
Why do repositories exist?
Repositories came about with the rise of Open Access. Gold Open Access is content that is made freely available on publication, while green Open Access refers to the author’s accepted manuscript of content that is published in subscription journals (and so hidden behind a payment barrier).
How do you find a repository?
That’s usually quite simple! Just search for the name of the institution and the word “repository” (which seems to be used in many countries).
Don’t repositories just duplicate the content in Web of Science and Scopus?
They don’t overlap completely. Some content (see below) is available in repositories but not in the big commercial aggregations, and some subscription content is not available from repositories. If you want to be sure you have searched for everything you need to look both in commercial and institutional aggregated collections.
Why don’t we hear about repositories?
Because, for the most part, we don’t access them directly; there is no need to look for content in an individual repository. Many aggregators collect all the content in repositories, and for finding articles, tools such as Unpaywall can be used with Google Scholar to locate a free version of an article, so the user may never know that the version they are reading is held in a specific repository collection. That doesn’t mean repositories don’t need to exist!
Aren’t repositories just inferior content?
Not really. Some changes may be made to the final published version, but a comparison of the author’s accepted manuscript (AAM) and the version of record (VOR) usually reveals little if any variation.
Why do repositories need to be managed?
Because researchers don’t, typically, bother too much about making their content available to the wider community. For many authors, sadly, publication is sufficient; they aren’t bothered if the article isn’t widely available. The “publish or perish” culture certainly exists, but to be precise, in this case, it’s “publish but don’t worry too much about availability”.
Why do repositories cost so much to run?
The smallest repositories have just one manager, but even one staff post that is a substantial commitment for any organisation. Without the repository staff, most repositories would struggle to have any content. Although institutions do everything possible to make researchers aware of the repository, it is never mandatory for a researcher to upload their paper, and many (most?) researchers would never do it unless prompted. Some institutions don’t have the resources to prompt their researchers, which means that content may never be widely available.
Secondly, the metadata is usually insufficient. The author’s accepted manuscript will typically be copy-edited, but often lacks some of the vital metadata that enables the published article to be found, for example, a DOI (a unique ID for the article), the author’s ORCID ID (which enables the author to be disambiguated), and the date when the manuscript was made available from the repository (usually called the date of deposit, although it might not be the same date that the article was uploaded to the repository).
Thirdly, the contents of the repository are typically in PDF format. While there are a number of machine-based tools to extract metadata from PDFs, this is not an exact science. The machine makes mistakes, and the correction process takes time.
Fourthly, for countries like the UK, deposit of government-funded articles in a repository is mandatory. To prove it has been carried out (an operation called the UK REF) is an exercise that costs millions of pounds each time it is carried out. Most of that cost is borne by the repository.
Do institutional repositories collect all content created by the institution?
Sadly, no. Much of the non-article content in journals, such as reviews and editorials, remains closed access, even though it forms an essential component of the scholarly process, particularly for the humanities.
Why do repository staff look for articles?
Many repository staff try to find everything published by a member of their institution, so they can remind the researcher to deposit a copy in the repository. This work is time-consuming. In countries such as the UK, depositing articles is a condition of funding by the UK government via UKRI.
Do repositories hold anything other than journal articles?
At the beginning, the goal was mainly to make articles available. Today, however, the scope has broadened enormously, covering:
- Theses and dissertations – most of which would never be visible if it were not for the repository
- Content unique to the institution, such as student newspapers, or other kinds of output that is not published in the normal channels
- Books, mainly monographs, which are increasingly being captured in the repository.
Could repositories be run more efficiently?
This is a big question, and something I will look at in future posts.
The biggest cost item for a repository is the staff, and the highest proportion of time spent by the staff is searching for and tagging content for the repository. The tragedy is that the published version, with all the relevant metadata cleanly coded in the correct fields, exists somewhere – and yet it is not used by the repository when they build their collection. That metadata is painstakingly input by the researcher, or more frequently, added by the repository staff.
This means, in other words, there is a cost of providing green Open Access. Institutions bear that cost, even if that cost is not always noticed, since it appears in the accounts as a staff cost rather than a cost per article.
What is more, if an article is retracted, there is an additional cost for the institution to keep track of the retraction and to take down the article or to issue a correction. It is an essential activity, but it costs money.
Are repositories changing?
For several years, repositories remained fairly static, but today there is something of a rebirth in the repository world. Many institutions are now storing the data for an article, as well as the article. But most importantly, the increasing use of machine-based processes for uploading and retrieval offers substantial savings for the repository team. Content can be checked for missing metadata, and that metadata can increasingly be found automatically, without the need for humans to find or copy content. However, this means an increasing technical requirement for the repository team, and the repository world is becoming an increasingly machine-based one. It requires different skills, and a different approach.
Finally, the use of APIs to find content on a large scale without having to look it up by hand is steadily increasing. Text and data mining has existed as a small-scale specialised activity in institutions for many years, but the use of a corpus for building AI tools is steadily increasing, and the need for better metadata, and for more sophisticated APIs, is therefore greater.
I’m sure I have missed some vital aspects of repositories. If you spot anything missing, please let me know!
Leave a Reply