A mandate for Open Access: The University of Liège (ULg) and ULg Library

The University of Liège has an exemplary Open Access mandate. It inspired a unique collaboration among researchers, university management and library. In this guest post, François Renaville, systems librarian at the ULg Library, explains how. ULg's institutional repository ORBi is indexed in Primo Central. For more information about institutional repositories in Primo Central click here.

Open access to academic output has been for some years now in the center of scholarly communication at the University of Liège (ULg). In 2007, our Administrative Board adopted the Immediate-Deposit & Optional-Access (IDOA) mandate and took the radical step of making it compulsory for our researchers to add references for all their publications and academic conferences to the institutional repository, backdated to 2002, and to deposit the full electronic version of all the articles that they have published since 2002.  

A reference to a scholarly publication has to be placed in the repository as soon as the publication has been accepted by the publisher or as soon as the document is considered to be complete. The obligation to deposit the full text of documents in the repository is relevant only for articles published in journals, but it is in no way exclusive: any other type of publication (book chapters, dissertations, reports, conference presentations, lectures for the general public, posters, course notes...) can be deposited as well.  Authors grant open access to the full text of the documents if the publishers’ terms allow them to do so. In cases where embargoes or other restrictions are imposed by a publisher, readers can still request as a print copy from the author directly from the repository.

The reasons behind this policy are simple. Firstly, University management needs to track its output. Bernard Rentier, Rector of the University of Liège, explained in his presentation at the MedOANet European workshop 2013: “If you don’t know what the scientific production of your university is, you have a problem.” But with an institutional repository, the institution can derive statistics for its output. Secondly, researchers are “not supposed to work for no readers,” as Thierry van Cutsem, Research Director FNRS and Adjunct Professor ULg, stated in an interview recorded in the video below. Bernard Rentier explains in his MedOANet presentation that scientists who are satisfied with being able to claim that they have published in a prestigious journal—even though it is only accessible for users in universities who can afford it—are completely missing the point. The point is that you want everybody to read your scientific output, to understand it, and to reuse it.

The ULg Library is playing a key role in supporting the seamless adoption of the IDOA institutional mandate. The library is tasked with the provision of an institutional repository and the support of the institution’s faculty and researchers depositing their publications. In November 2008, the library launched its DSpace-based institutional repository ORBi within the collaborative framework of the Belgian French-speaking university library consortium (BICfB). A few months later, the ULg’s mandate was further advanced by a policy advising that evaluations, appointments, promotions, and budget allocations decisions will take into account only references that are archived in ORBi.

 

Currently, ORBi contains more than 94,700 references, of which about 57,800 (61%) have a full text (29,900 of them [52%] with an OA full text), and gets more than 2,400 downloads a day (spiders excluded).

Besides the endorsement of the strong institutional policy, other factors played a major role in the success of ORBi. One of them is the user-oriented improvement of the repository: more specifically, we consider that referencing and dissemination is fundamental to reach a real increase of the visibility of the authors’ academic production. On that point, ULg Library’s efforts are very effective. For example, once a reference has been archived in ORBi, it is searchable in Google within an hour of archiving. ORBi is also harvested by other search engines and archives, e.g., Google Scholar, OpenAIRE, Base, Driver, Isidore and, of course, the Primo Central Index.

ULg conducted a number of interviews with researchers, recorded in the video below, confirming that they are well aware of the benefits. In addition to reaching a very large readership very quickly, researchers see their impact sooner. Christian Hanzen, Professor at the Faculty of Veterinary Medicine, states: “When I effectively consult the number of my downloads and I see where they originate from, I realize the impact I can have.” 

For more information refer to the two videos below, featuring interviews with researchers at Liège, and in the second, a presentation by Bernard Rentier explaining the institutional mandate. 

For more information about registering institutional repositories for Primo Central click here. 

Tweet

Challenges and Transformation at ALA Midwinter

I had the pleasure of attending the American Library Association Midwinter Meeting in Seattle last month and was impressed with the incredible interest librarians expressed in addressing the challenges they face while transforming their processes and expanding their services.  We all know the challenges – fewer staff, more and more resources (particularly electronic), stagnant or decreasing collection budgets, increased focus on demonstrating value and relevancy to the organizational mission and goals, and more.  These challenges, or at least some of them, are now widely acknowledged and it is encouraging to see the discussion shift from recognition to problem solving.  I sensed a real focus on transformation at this year’s meeting: transformation of services through more collaboration and cooperation, transformation of traditional cataloging to emphasize metadata services that enhance discovery, and transformation of workflows to take advantage of next-generation unified systems.

This increased emphasis on collaboration and cooperation was clearly demonstrated when over 150 people crowded the Orbis Cascade Alliance and Ex Libris session entitled “Unlocking Opportunities for Collaboration” presented by John Helmer and Lynn Chmelir from Orbis Cascade, and Asaf Kline from Ex Libris.  The Alliance, which has a strong history of cooperation across its vast membership, discussed its decision to move from legacy systems with limited bandwidth and lagging functionality, to a next generation infrastructure based on Primo and Alma.  They envision better services for students, faculty and staff, improved resource sharing, and new opportunities for collaborative technical services enabled by Alma. 

The Orbis vision of “one collection” was echoed in an ALA session I moderated with Roger Brisson from Boston University, an Alma Early Adopter.  Roger’s session, which was presented to another overflowing crowd, focused on how Alma has been a catalyst to fundamental changes to the technical services environment at BU and how Primo, along with the library’s new website, is further demonstrating the value of the library’s services. For details on Roger’s presentation, please see the coverage in Library Journal’s Digital Shift.

All in all, it was encouraging and rewarding to see the vision of Alma and Primo being realized by such forward thinking librarians.

Susan M. Stearns

vice president of strategic partnerships, Ex Libris 

susan.stearns@exlibrisgroup.com
Tweet

Institutional Repositories in Primo Central: Registration is now open

Institutional repositories contain lots of treasures including rare or otherwise unpublished material and articles that scholars self-archive, often as part of their institution’s open access mandate. But it can be hard to discover this material unless users know exactly where to look. We recently released a new service for institutional- and open access- repositories that will simplify the process of allowing their content to be indexed in Primo Central. Primo Central is a great showcase for this material; it makes it globally discoverable and provides exposure to a large user community of now over 1300 institutions.

The new registration service is open to all institutions, not only to Primo customers (read more here). The goal is to enable users at Primo institutions to discover all available material equally. The Primo Central index already contains a number of repositories such as Harvard’s DASH.

Here are a few more thoughts about institutional repositories. While open access publication in general plays a key role in providing free access to publicly funded research, institutional mandates can play a key role in accelerating self-archiving in institutional repositories. Yassine Gargouri, Stevan Harnad et al note in their study Green and Gold Open Access Percentages and Growth, by Discipline: “Mandates almost immediately triple the baseline Green OA self-archiving rate”. The Registry of Open Access Repositories Mandatory Archiving Policies (ROARMAP) lists over 250 mandates while the Directory of Open Access Repositories - OpenDOAR lists over 2200 repositories worldwide – both with a steady increase.

There are a number of publications that help institutions to shape their repository strategy. Peter Murray-Rust’s blog post Criteria for successful Repositories gives an excellent overview of the issues to consider when establishing an institutional repository. Harvard University publishes in-depth advice on their Wiki Good practices for university open access policies. In their very interesting article, Institutional Repositories: Exploration of Costs and Value, C. Sean Burns, Amy Lana and John M. Budd discuss the results of a recent survey and highlight dependencies on whether institutions mediate submissions or allow un-mediated deposits, the software type used, and what other services are being offered by the institution.

For institutional repositories to be successful, authors must have clear incentives to deposit their material there. Discoverability (and consequently better chances of being cited) is a powerful motivator and Primo Central makes IR materials as discoverable as other content. The emergence and uptake of alternative usage-based metrics—as described in my altmetrics post—emphasizes the need for exposure to achieve impact scores that can help with tenure and grants. Finally, open access publishing in general is becoming increasingly attractive for authors as arguments against it are starting to crumble.

Here are two stories that illustrate this change: Firstly Mike Taylor’s article Hiding your research behind a paywall is immoral. (Chris Chamber’s response Those who publish research behind paywalls are victims not perpetrators  is an equally interesting read). Secondly, projects based on self-archived content have begun to emerge, such as the Episciences Project, to create a platform for arXiv overlay journals, which is discussed in Tim Gowers’ blog post Why I’ve also joined the good guys. Whichever open access route authors take – gold or green – exposure and discoverability are important to all of them.

The institutional repository registration service for Primo Central is part of a wider Ex Libris initiative to support open access. In addition to adding more open access material to our indexes we are also improving access to open access articles in subscription (hybrid) journals (discussed in an earlier blog post) and we will continue to enhance this initiative.

Tweet

My favourite EL Commons code extensions of 2012

It is over four years since we introduced the Ex Libris EL Commons collaborative platform. Customers and staff members use the CodeShare section of EL Commons to publish the extensions they build using the Ex Libris systems’ open interfaces. Here are three of my favourite code contributions of 2012:

The Primo Result Extender Aggregator Service (PREAS) was created by Karsten Kryger Hansen and Kasper Løvschall from Aalborg University. It allows libraries to enrich Primo search results with supplementary information such as matching full text in Google Books, related information in Wikipedia, and the availability of article feeds from JournalTOC for specific journal titles.

Lucas van Schaik from Leiden University created another nice extension with a program that adds Google-like pagination to the Primo result pages.

Both contributions did not just achieve high ratings by fellow developers, but were also very popular among Primo users; they were each downloaded more than a hundred times.

Last but not least: Shortly before Christmas, Ido Peled, product manager at Ex Libris, added a code extension that adds altmetric scores from altmetric.com to the Primo research results, where they are available. It captured a lot of attention and has been downloaded over 70 times in the short time since it was published.

These three favourites are just a selection of many great contributions. Ex Libris customers can view many more on the EL Commons website. Sharing code extensions and trying out new ideas based on the open interfaces of Ex Libris systems adds a lot of value for users and allows institutions to gain from each other’s expertise. Supporting EL Commons is part of the Ex Libris open platform strategy.

We are looking forward to a great 2013 with many more innovative ideas!

Tweet

Altmetrics on Primo

We have just added a free Primo code extension to the CodeShare area in EL Commons, the Ex Libris collaborative platform. It enables libraries using Primo to embed altmetrics from altmetric.com into the search results. I am very excited about this new feature. Read more about my take on altmetrics below the picture.

The altmetric.com score on Primo:

Let’s start with a brief description for those who are not so familiar with altmetrics. Altmetrics stands for alternative metrics. Generally speaking, these metrics are counts of different usage events for scholarly articles, that are taken from various sites. Such event counts, for a specific article, may include the number of times it was cited  from Scopus, CrossRef and Wikipedia; the number of bookmarks on Mendeley and CiteuLike; the number of downloads of PDF or HTML views on a publisher’s website; and the number of comments on Twitter. A splendid example of such metrics can be found on PLoS where every article shows a metrics tab. Further examples and details can be found in The Altmetrics Collection by Jason Priem, Paul Groth, Dario Taraborelli (2012) and on the ImpactStory website

To better understand the significance of altmetrics, it’s important to look at the wider context. Scholarly communication is undergoing dramatic changes in our web-driven world. Open access policies are increasingly being endorsed by institutions and funding agencies, and there is a growing demand for indicators showing the impact of research output.  Social media play an ever more important role in how research is disseminated. Through social networks, users post comments, reviews, and conversations, and provide feedback immediately after or even before publication. Stacy Konkiel and Bob Noel state in their excellent presentation Altmetrics and Librarians: How Changes in Scholarly Communication will affect our Profession (2012) that the “feedback loop is shortened, accelerating research”.  Other interesting material evaluating use of social media by scholars include Carolyn Hank’s The Scholar Blogs of Today, Tomorrow: Practices and Perceptions of Value, Impact and Stewardship (2012) and John Conway’s Blogs, Twitter, Wikis and other on-linetools (2011). 

With the rapidly multiplying numbers of available scholarly publications, many of which are open access, readers need ways that help with sifting this material. Traditionally, the Impact Factor played a key role in researchers’ decisions about what to read and cite. But as Cameron Neylon and Shirley Wu state in their article Article-Level Metrics and the Evolution of Scientific Impact “… the impact factor … is simply not designed to capture qualities of individual papers.” With the rise of the web, measures based on usage and social media have become available that capture impact in a different and more immediate way. Such measures are derived from the usage by a far bigger community than the group the Impact Factor is based on (which only relies on scholars who publish and thus cite) and are applicable on article level. The impact of research and scholarly publications is difficult to define since it can take different shapes and forms. The UK Research Excellence Framework (REF) states in their 2011 report Decisions on assessing research impact that “The impact element will include all kinds of social, economic and cultural benefits and impacts beyond academia”. In short, measuring, qualifying and quantifying impact and using this for recognition (tenure, grants and promotions) and selection of material is very important but equally a difficult and multifaceted task.

Since articles are used and reviewed in many different places, altmetrics are also created at many different sites. A key benefit of looking at a variety of altmetrics is that they can provide a picture of usage, beyond just an individual measure. Also, they start appearing immediately with or before formal publication and the values they provide are usually quite transparent. Altmetrics cannot replace any qualitative peer review but often point to it, for example in the form of comments on social media or reviews on F1000. Altmetrics enhance rather than replace other evaluation methods, and provide a rich picture as is suitable for today’s rich and diverse environment.  Jonathan Eisen, Prof. at UC Davis, gives an interesting example for the use of altmetrics on his blog post Playing with Impact Story to look at Alt Metrics for my papers, data, etc.

Altmetrics do have their weaknesses: Firstly usage data is subject to gaming, that is, manipulation by an individual or a machine. There are preventive measures sites can take but it cannot entirely be ruled out; however, since usage comes from different places, it is very unlikely that all numbers are manipulated in the same way. In addition, usage is distributed and not all places provide altmetrics. Hence, lack of altmetrics or low numbers do not necessarily mean that an article is not used. And finally, usage is subject to interpretation: the intention of a bookmark, tweet, html or pdf view cannot be measured at this time; they do however indicate explicit interest for an item. Even with these caveats there is no question in my mind that altmetrics provide a significant enrichment of the user experience. In future, we will add our own metrics from the bX Usage-Based Services to further enhance the usage picture for an article with the usage data that we gather from hundreds of institutions around the world.

Tweet

From the bX data lab: Topic visualization and usage data

Experimenting and testing is an important part of our bX work. In our bX data lab, we work with data samples to check the quality of our services, improve them and evaluate new ideas. I recently conducted a small experiment relating to topic popularity and relationships between topics and subtopics, an ideal subject for trying out some visualization software, and here are the results.

First I took the 20 most used articles from each of the medical categories in the monthly Hot Articles indexes from March to July 2012, and extracted the articles’ medical subject headings (MeSH). Not all the articles had subject headings and my final sample consisted of about 300 articles. I then checked for the most popular topic which turned out to be “Obesity”. From all articles with this topic, I extracted all main subject headings and used Gource and ScreenR to create this little video that shows a visualization of the most popular aspects of Obesity at present.

Why Gource? I watched Tony Hirst’s visualization of OpenURL referrals, made with Gource, some time ago, and really liked the idea. Gource is an open-source software available from the Google code web page.  It is actually a software version-control visualization tool and was not meant for my purpose, but it is also quite easy to tweak it to do (almost) what I wanted. Gource requires a very simple input file in txt format that can be created from a logfile or from a spread sheet. The format for the input file is:

<unix timestamp>|<User>|<Added or modified (A or M)>|<directory/file name>

I had to repurpose the input parameters since my little project is about popular topic associations. I took as the user a number. For my purpose, it was not really important what number, so I took a bX internal article number and - instead of a directory and file name - a heading and subheading. The latter I had to tweak because Gource tends to show all labels on top of each other if there are no directories to separate them, so I grouped the headings and added directories for each group. My visualization does not show how the topic changes over time; rather, it shows the topic associations that are popular at present. Therefore I re-purposed the timestamp and used it to trigger the directories to appear sequentially, rather than all at the same time, for easier viewing. And that’s about it. The Gource commands are easy enough and can be run from command line on a windows computer. There is a nice help screen available by just running gource –H. For producing the video, I used ScreenR to capture it from the screen.

My little experiment only includes data from a few months and for only one topic. But the experiment does show that analysing topics based on popularity can be a very interesting exercise, if, for example, it relates to usage data for a much longer time period. Key topics of interest and the items that users view most will change over time, and a good visualization tool can show how interest in a topic and in its associations with other topics, change and evolve. Visualization tools can also uncover trends and correlations and reveal 'hot' topics in the past and at present.

A good example for uncovering connections and trends is Google Correlate, which shows correlations between different search keywords and real-life data. Another example is the Google Flu map that estimates where the next flu outbreak will occur – all based on usage data. Unlike the Google data, our bX data is purely focussed on the use of scholarly material and can therefore identify trends in scholarly discourse. Fascinating stuff!

Tweet