Tamar Sadeh, Director of Marketing, Ex Libris
At the end of 2009, only six months after the development of Primo Central was announced by Ex Libris, this mega-aggregate of scholarly materials went from theory to practice. During the end-of-year holidays, the Ex Libris implementation team began activating Primo Central at Ex Libris library partner sites for testing. A new component of the Primo® discovery and delivery system, Primo Central enables institutions to expand the Primo search scope to include not only local collections but also global scholarly materials. Results originating from Primo Central are blended with local results, yielding one relevance-ranked result list (figure 1).
Figure 1. A blended result list at Vanderbilt University. Click to enlarge
The University of New South Wales was the first partner institution to start testing Primo Central, at the beginning of January, 2010. At the time of this writing (the end of January), the following libraries have begun testing Primo Central:
- University of New South Wales (Australia)
- Vanderbilt University (U.S.)
- Oxford University (U.K.)
- Catholic University of Leuven (Belgium)
- Cooperative Library Network Berlin-Brandenburg (KOBV consortium), which includes Humboldt University of Berlin, the Free University of Berlin, the University of Mannheim, and the Technical University Berlin (all in Germany)
- Yonsei University (South Korea)
Next in line are the following institutions:
- Brigham Young University (U.S.)
- Michiana Academic Library Consortium, which includes the University of Notre Dame, Saint Mary's College, Bethel College, and Holy Cross College (all in the U.S.)
By the second week of February, we expect all Primo Central library partners to begin testing. MetaLib® Primo Central partners will start testing Primo Central as a MetaLib resource at the end of March. In mid-2010, Primo Central will be available to all Primo customers; it will be activated gradually at MetaLib customer sites during the third quarter of 2010.
One of the greatest benefits of the Primo system is that it serves as a single point of entry from which users can search in all library collections. Libraries, in presenting their own collections—library catalogs, digital repositories, course materials, and more—are taking advantage of the capability of Primo to enhance library data through the system’s integration with data-related library and non-library services such as those offered by Amazon, Syndetic Solutions, and OCLC WorldCat®. In addition, Primo customers have been demonstrating a high level of innovation in customizing the system, integrating it with other library and institutional systems and services, and developing mobile and other alternative interfaces for users. Indeed, the statistics bear out the popularity of this system. At New York University, the average number of scholarly search sessions per day increased from 4,000 before Primo was implemented in September 2008 to 10,200 a year later, while the average duration of a session decreased from 15 minutes to six minutes during that period. Yonsei University in Seoul reports similar statistics. Clearly, Primo users are finding what they need much faster.
Primo also enables libraries to extend the scope of the collections they offer to their users through Primo by harvesting metadata from other collections and presenting those collections in a way that is transparent to end users. For example, Boston College has added to its local Primo index more than 170,000 records from the University of Michigan’s MBooks collection, and the University of Minnesota, a HathiTrust member, has added over 600,000 HathiTrust records to the university library’s local Primo index.
It was only a matter of time until Ex Libris started leveraging one of the fundamental elements of the Primo architecture—the capability to search in distributed index segments—so that the search scope of Primo could be extended even further and in a way that accommodates all Primo libraries. Because of the global nature of academic research, it is clear that library users, regardless of their geographic location and institutional affiliation, are interested in the same global resources. The availability of these global resources along with local resources renders the discovery process streamlined and rapid for end users. However, harvesting such global collections and adding them to a local Primo index is not a feasible solution for libraries. The Primo Central index, which Ex Libris maintains and makes accessible to all Primo libraries, offers an efficient, cost-effective solution. Mark Dehmlow, of the University of Notre Dame—a Primo Central library partner—explains:
Given that library budgets are finite, library managers will always need to set priorities for the types of projects we engage in. It is increasingly clear that we must use our personnel and fiscal resources to add value that our governing institutions and users can understand and appreciate, leveraging what makes us stand out from other institutions, e.g., digitizing rare books and special collections, creating metadata for unique items that haven't been cataloged anywhere, and customizing their systems to meet their patrons' specific needs.
Medium-to-large institutions will likely continue to locally host many of their core services and systems for the foreseeable future, but there’s a problem of scalability for libraries to host and manage all of the data that are useful to us. In my mind, Primo Central is the "right scale" software as a service (SaaS) for aggregating the growing body of licensed and free content. While the advantage of hosting this content on a single infrastructure is obvious from the perspective of search efficiency, the process for acquiring and managing this data locally would be arduous at best for libraries and, more importantly, would divert our valuable resources away from projects that can add the most value for our institutions. My thinking is that it is worth the cost to have someone else manage the logistics for this type of data—the contracts to get the data, the normalization and loading of the data, and the high-availability provisioning of access. This way, libraries can continue to put our efforts into making our unique collections more accessible, delivering core services, and creating the best library experience for our users.
Because Primo Central fits into the Primo architecture as a Primo index segment, the blending of Primo Central results with local Primo results is straightforward and yields the same rapid response time as a local Primo implementation. Furthermore, in the soon-to-be-released Primo 3.0 system, a library will be able to use the Primo administration module to tweak the relevance-ranking algorithm so that the blending of local results with Primo Central results will match the library’s policy regarding the prominence of its materials in the result list. Of course, libraries will continue to have control over the exact search scope, the user interface, and the manner in which results are displayed (regardless of the results’ origin).
The creation of Primo Central has been a great experience for the development, implementation, and data services teams at Ex Libris. Many team members have worked together to obtain data from information providers, normalize the data, and map it for the Primo index structure; to develop blending and relevance-ranking mechanisms; and to build an environment that supports hundreds of millions of records and a very large number of Primo and MetaLib users while retaining the excellent response time and relevance ranking that characterize Primo. With the goal of maintaining a robust infrastructure, facilitating access from all over the world, and addressing peak periods of usage, Ex Libris is hosting Primo Central in a cloud computing environment.
And, indeed, so far so good: as libraries are beginning to work with Primo Central, we are witnessing a smooth transition to an extremely broad information landscape that is growing at a very rapid pace. Over 200 million records are currently being added to Primo Central as a result of agreements with information providers, and many more records are expected to be added within the next few months.
 E-mail communication from Corey Harper at NYU to the NGC4LIB listserv, October 10, 2009
 Figures presented at the Ex Libris user group meeting in Seoul on November 13, 2009
 A collaborative of 26 universities that aim to establish a repository where they can archive and share their digitized collections (http://www.hathitrust.org/)