May 24, 2018

The Lure of Linking

Link resolvers are essential to getting optimal usage of electronic content

One of the most innovative and revolutionary library services to evolve in the Internet era is reference linking, the ability to transmit bibliographic data through hypertext links and to connect users with the full richness of electronic collections with ease. These collections can include academic research journals, general interest magazines, e-books, bibliographic indexes, and other reference works – as well as the sources and services associated with those items. Reference linking lets users move from an abstract from one publisher to the full-text document in another vendor’s database, from a bibliography in an article to the library catalog, from one database to another, or from a journal article to a web site.

Reference linking is necessary because library and information users today expect to move seamlessly among library content and information on the Internet. Libraries present users with disparate databases, different user interfaces, various searching capabilities, and changing institutional subscriptions. Reference linking is largely succeeding in removing these barriers. For more about how link resolvers work, see “Link Resolver 101 ,” p. 33.

Resolving appropriate copy

By 1998, context-sensitive linking was a topic at library conferences. Partial solutions had been proposed, but none could solve the most vexing of problems: the appropriate-copy problem. This occurs when linking from a third-party database to a resource (say, a journal) of which one has multiple copies. Each of these copies is governed by distinct access policies. Ideally, the links to this resource should resolve to the copy that is appropriate to the particular user and circumstance (the “context”). Problems stemming from owning multiple access licenses to the same resource may seem limited to only the richest of institutions (thus the nickname “the Harvard problem”), but they were a stumbling block to development.

At this crucial time, Herbert Van de Sompel, then head of library automation at the University of Ghent in Belgium, proposed a link resolver managed by the library. Because the library knows its collections, its policies, and its users, a library-managed link resolver makes context-sensitive linking possible. In collaboration with Patrick Hochstenbach, also then of the University of Ghent, Van de Sompel proved the viability of the concept by implementing the SFX (named for special effects) linking server. It was quickly recognized as the silver-bullet solution. In October 1999, Van de Sompel along with Los Alamos National Labratory’s Rick Luce and Cornell University’s Paul Ginsparg organized a historic meeting, the Santa Fe convention. The meeting, best known for the Open Archives Initiative, also put other digital library projects including SFX on the national stage.

The Caltech experience

In early 2000, Ex Libris, known for its ALEPH Integrated Library Catalog, acquired the rights to the software and beta tested with a small number of U.S. libraries. The beta phase was designed to test the viability of the reference linking concept and extend beyond early experiments at Ghent with SilverPlatter databases. The California Institute of Technology (Caltech) Library System got involved because Eric Van de Velde returned from the Santa Fe convention convined that Caltech needed SFX.

Staff attended beta training at the Ex Libris offices in Boston in 2000. Soon after, Betsy Coles, manager of digital library systems, wrote the code that made reference linking possible from the ISI Web of Science databases that the library had locally loaded. John MacDonald, acquisitions librarian, uploaded the information about the library’s journal holdings to the resolver’s database (now known as the Knowledgebase).

Thereafter, Caltech focused on getting more information providers to “enable” their databases to work with SFX and to create linking syntaxes to other information resources. This was not always easy. Many commercial providers met the request with either a quick denial or a request for additional subscription payments for added services. In the best cases, companies simply asked for information and clarification of what they needed to “OpenURL enable” their databases.

Going live

After a year of testing and development, the resolver was released to the campus community in April 2001. Initially, it allowed users to link from an entry in the Web of Science to outside services, including links to full – text articles in other databases, interlibrary loan services, and search engines on the web. We opted for a “soft” launch, with minimal advertising to the community. The release of SFX made Caltech the first academic research institution in the United States to have a resolver active in a production environment

Caltech patrons used the SFX links without delay and with few problems. In fact, the first feedback we received was an inquiry about adding additional full-text subscriptions. The second was to report a problem the user had with the browser. Soon after, we began receiving love letters from our patrons praising the new linking technology. Remarkable, since we rarely hear from patrons unless something is wrong.

Content providers get on board

We hesitated about releasing the resolver without a critical mass of both Sources and Targets. We feared that users would only associate the resource with links to full-text articles from the Web of Science. By April 2001, only EBSCO and Institute of Physics, among our vendors, had OpenURL in production. Our patience was running out, and we decided to release the service anyway.

After our release to the campus, other providers began to OpenURL enable their products. By the end of 2003 most of our vendors were enabled, with the exception of Innovative Interfaces, our integrated library system. Innovative chose to require customers to purchase its competing OpenURL resolver, WebBridge, before allowing OpenURL linking from the catalog, and it remains the only major resource at Caltech that is not a Source.

Caltech’s Knowledgebase currently has 34 information resources active as Sources and 95 active as Targets, encompassing nearly 6000 electronic books and 4000 full-text electronic journals. A smashing success with faculty and students, it receives nearly 10,000 full-text downloads per month, over a quarter-million articles over the past 28 months.

Other vendors arrive

ExLibris’s SFX resolver was the first to the market, but many other vendors now offer competition. Some are standalone products used solely for resolving OpenURLs while others are integrated into OPACs, proprietary databases, or other services. In addition, there are individual efforts underway to build resolvers locally using OpenSource software. The major standalone resolvers are SFX from ExLibris, Endeavor’s LinkFinderPlus, Journal Finder from the University of North Carolina – Greensboro, Gold Rush from CARL, SirsiResolver from Sirsi, Fretwell-Downing’s OL², 1Cate from Openly Informatics, Serials Solutions’s ArticleLinker, and LinkSource from EBSCO.

Many of these vendors offer an ASP option, which means that the resolver is hosted by the vendor, with support in establishing the “opening-day collection” for the Knowledgebase and assistance with its continuing maintenance. Many also offer the products as software with annual maintenance fees. This allows the library to run the resolver on its own hardware and retain local control.

While most OPACs are now OpenURL enabled as both Sources and Targets, some go a step further and integrate an OpenURL resolver directly into their products. Both Innovative with its WebBridge product and Dynix with Horizon Link offer this option.

In addition, content providers are building OpenURL resolvers into their proprietary database systems. Companies such as H.W. Wilson and Cambridge Scientific Abstracts have OpenURL links built into their database interfaces. This allows users who do not have OpenURL resolvers to still use OpenURL technology to build additional services, as predefined by the database vendors themselves.

For help determining which product is right for your library, see “Considering a Resolver?” on p. 34. But some libraries might want to build their own. Several libraries with the technical staff and motivation have opted for this solution. It offers a few advantages over commercial resolvers, especially greater flexibility. “First of all, building our own resolver enables us to load our Knowledgebase to suit our own situation,” says Mark Dahl, library technology coordinator at Lewis & Clark College, Portland, OR. “It also allows us to share our resolver with campus libraries.”

The resolver experience

As an initial beta-tester, Caltech had a unique experience. More recent implementers, such as the library at Macalester College, a small liberal arts school in St. Paul, are more typical. Angi Faiks, collection management team leader at Macalester, has found linking technology enhances the scholarly process. “We can say with certainty that you don’t have to be a large, research-focused, graduate-level institution to take advantage of what OpenURL has to offer,” says Faiks. Macalester launched SFX over two years ago using the ASP model.

The service has been popular, with over 1000 SFX requests made daily in 2003. Faiks reports that they have greatly reduced the number of interlibrary loan requests, a side benefit to connecting to resources in disparate databases. “Plus, by populating our loan request forms with accurate citation information we have saved a great deal of time,” says Faiks. Macalester manages its Knowledgebase with a single staff member who had SFX maintenance added to previous duties. “Even in tough budget times, this is money well spent,” adds Faiks.

The University of Richmond, VA, implemented Endeavor’s LinkFinderPlus in fall 2002 after a fast-track implementation of just under a month. Like Faiks, Rachel Frick, Richmond’s head of bibliographic access services, believes their linking tool “increases the value of our e-resources by increasing their usage.” Frick also has no concerns about the flexibility. “We have customized our interface,” she says. “We also are in the process of adding a search for author or title to run through the information in OAIster [a metadata harvester for depositories in the Open Archives intitiative]. It is this type of search that brings together the power of Open URL with the powerful idea of open access materials.”

Not just academics

Customers for link resolvers are as diverse as theological schools, historical societies, and special libraries. Susan Scheiberg, assistant library director, RAND Corporation, says that their implementation of Serials Solutions’ ArticleLinker resolver “revolutionized our library service in about 15 minutes.” Scheiberg was initially attracted to the product because she was an existing Serials Solutions customer and because of the product’s low cost. With the ArticleLinker up and running, Scheiberg feels that staff and patrons can better identify the disparate databases that the library purchases but also maximize use of full-text databases, online journals, and abstracting and indexing databases.

Even with this good news, public librarians have been slow to implement link resolvers, something that baffles most vendors. It is crucial that public libraries use link resolvers, argues Peter McCracken, cofounder of Serials Solutions, especially since public libraries often offer large journal collections provided through statewide contracts. Perhaps public libraries have been put off by the technological investment, although McCracken points out that an ASP solution is perfect for small and medium-sized public libraries.

Future of linking

It is safe to guess that in the not-too-distant future every library with a digital collection will provide an OpenURL resolver for its users. As a rule, the OpenURL resolver becomes an essential part of the digital library infrastructure within months of deployment. Users depend on it, and librarians insist on its reliable availability. The future of resolvers in digital libraries is secure, at least until a substitute technology comes along.

The question that remains is whether this technology will find its niche outside of library applications. The NISO (National Information Standards Organization) standardization committee certainly thought so when it transformed the basic OpenURL idea into the OpenURL Framework last year. The Framework is agnostic as to application domain, which means it can be applied to other industries. For example, the real estate industry could define metadata and identifiers for houses, realtors, buyers, sellers, various contracts, and more – creating a whole new set of applications.

Van de Sompel, now at the Research Library at the Los Alamos National Laboratory, puts it as a challenge to NISO. “The committee has made sure that this generic specification is readily implementable in the community in which OpenURL emerged. However, whether this enormous endeavor will succeed will depend on whether other communities discover and adopt the standard. I feel that NISO has an important role…promoting a NISO specification beyond the typical NISO constituency.” The OpenURL Framework standard is ready for a few enterprising individuals to use it to explore whole new worlds.

John McDonald ( is Acquisitions Librarian and Eric F. Van de Velde ( is Director of Library Information Technology, California Institute of Technology, Pasadena

Link Resolver 101

A student searching through PubMed finds a citation to an article she wants to read. But how to locate a copy? Researchers today are faced with questions that range from whether the journal is available online to whether their library has a subscription (either print or online), from wondering if the journal may be in an aggregated collection to whether (and how) a copy can be obtained through interlibrary loan.It is difficult to answer these questions through conventional web links. After all, how would PubMed know to which other services the student has access? It all depends on the student’s affiliation.Link resolver software brings together information about the cited resource, the user, and the library’s many subscriptions, policies, and services. For the software to work, the content providers must be willing to participate as Sources (databases or sites that can provide a link from a reference). To make linking more reliable, it helps if providers colloborate as Targets (the end results, usually a full-text article or a search for print holdings in the library’s catalog). Also, the library needs to load accurate data about its holding into the link resolver’s Knowledgebase.The link resolver becomes activated when the user clicks on a link or button (“Search for full text”) embedded in the user interface of PubMed (or other services). Using the OpenURL Framework, information, such as the metadata for that citation, is bundled together from the Source and sent to the resolver software that will process the data, comparing it to the Knoweldgebase.As a result, the student is then presented with a range of options for locating the article. If the library has an online subscription, the Target may link to the article itself or to the journal. Other options include the library’s print holding for that title, or interlibrary loan or document delivery options.

Considering a Link Resolver?

Librarians who want to add an OpenURL resolver to their library services should evaluate not only the product but how it meets their local needs. Harry Samuels, product manager for Endeavor’s LinkFinderPlus, has developed a list of things to consider.First, the link resolver must be OpenURL-compliant to work with the vast majority of information resources in the market today. It should also be available as a standalone product so that the library has maximum flexibility in choosing the best solution for each piece of the digital library puzzle.Decide whether you want to purchase a system that is run and managed locally or one that is a subscription service hosted by the vendor. A subscription service may be more feasible for customers with limited technical staff or hardware resources, while a locally run system gives the library maximum flexibility and control. Make sure that any subscription service selected is hosted at a professional hosting site with 24/7 support and redundant connectivity to the Internet. Locally run resolvers should be able to run on several different operating systems and use well-known, commercially viable components that can easily be supported.Carefully evaluate the product’s Knowledgebase and match your own collections to the Sources and Targets with which the resolver functions.Will you want to customize the product interface? Some products allow significant customization; others may not. Customization of both the interface and the product itself might be important for some libraries. You may want to restrict some services, empower other services, or develop new ones.Most importantly, consider a product’s overall price. Not all resolvers are affordable for all libraries.

Maker Workshop
In this two-week online course, you’ll create a maker program that aligns with your budget and community needs, with personal coaching from maker experts—from libraries and beyond—May 23 & June 6, 2018.