The program stems from activities of the International Association of Scholarly, Technical, and Medical Publishers (STM). Publishers participating currently are Academic Press, a Harcourt Science and Technology Company (http:// www.academicpress.com); the American Association for the Advancement of Science (publisher of Science) (http://www.sciencemag.org); the American Institute of Physics (AIP) (http://www.aip.org); the Association for Computing Machinery (ACM) (http://www.acm.org); Blackwell Science (http://bslnet.blackwell-science.com); Elsevier Science (http://www.elsevier.com); the Institute of Electrical and Electronics Engineers, Inc. (IEEE) (http://www.ieee.org); Kluwer Academic Publishers (a Wolters Kluwer Company) (http://www.wolters-kluwer.com); Nature (published by Macmillan Magazines, Ltd.) (http://www.nature.com); Oxford University Press (http://www.oup.co.uk); Springer-Verlag (http://link.springer-ny.com); and John Wiley & Sons, Inc. (http://www.wiley.com). They hope to expand the program to other primary publishers.

Linking has been available in some online systems for many years, but usually the links connect abstracts or citations in one database on a system to full-text in another file on the same service, e.g., Ovid's links between MEDLINE and its full-text medical journals. LEXIS-NEXIS has an elaborate system of linkages in its legal material that can reach inside articles and connect from footnotes or references, as does WESTLAW with its recently announced footnote linking for law review articles, but still only within databases on each proprietary system. Rarely have services used open system, HTML architecture. SilverPlatter Information's SilverLinker service does link from citations to full-text articles on publisher Web sites, as does the National Library of Medicine's PubMed service. However, the systems that use Web-based links have not offered access at the footnote or endnote connecting level. Nor have these systems succeeded in gaining approval from all the publishers their databases cover or even all the publishers with fully archived Web sites.

Experts agree that researchers and scholars want this level of service. Richard Lucier, head of the Digital Library at the University of California, considers this "a very important functionality of strategic market importance to publishers." From the very beginning of the Digital Library at UC, Lucier has insisted that licenses with publishers include specific written clauses that let UC staff link references in the databases on their servers (e.g., MEDLINE) directly to publisher Web-based articles, but they have not had the option to move from citations within full-text articles to other full-text articles—until now. Lucier considers this a very important step. John Regazzi, head of Elsevier ScienceDirect, looked upon the new initiative as recognition by publishers that they need to offer an "industrywide solution." As Regazzi stated, "People think in terms of publications, not publishers. They want a unified search system."

The STM's executive board developed the press release announcing this new service at the Frankfurt Book Fair earlier this year. Rumors persist that some of the motivation for the release stemmed from a desire to respond to the challenge posed by the National Institutes of Health's proposed E-Biomed (a k a E-Biosci and now PubMed Central) system, which would have constituted a major Web publishing alternative for medical research in its earliest design. Since Dr. Harold Varmus, director of the NIH, announced his controversial proposal, medical publishers have responded vigorously and usually negatively. However, other government agencies seem to have taken up the call, e.g., the Department of Energy's new PubScience service. With the U.S. government marching to the Web's tune, scholarly publishers—both commercial and learned societies—may have felt the need to demonstrate their own commitment to full Web service designed to integrate products to the satisfaction of readers.
 

Technicalities
Typically material will be located on different servers run by different publishers. Initially, the service should link approximately 3 million articles across thousands of journals, with over half a million more articles added each year. Technically, the service will operate around a central facility managed by an elected board and in cooperation with the International Digital Object Identifier (DOI) Foundation (http://www.doi.org).

Dr. Joel Baron, former publisher of the New England Journal of Medicine, has taken a leadership role in the new reference-linking project, working with the STM and on the board of the International DOI Foundation. He expects the "Central Facility," as they call the new service, to become an official DOI registry in time. Baron agrees that conditions of access/pricing will vary from publisher to publisher, but said it would be unrealistic to expect otherwise. "We would never get agreement from all of them. We're better served by doing what we can now," Baron said. In fact, he asserts that having many business models in place could provide real value for the industry and possibly consumers.

Baron expected the first opportunity users will get to see the new system could occur as early as December 10, when the STM Innovation Committee meets in London. That might also be the time when the service gets a new name—something more attractive than "Central Facility." At our deadline for this article, the short list for a new name was rumored to be down to two candidates.

The new service will use a limited set of metadata, allowing the journal content and links to remain distributed at publishers' sites. The prototype for the service was developed by John Wiley & Sons and Academic Press, in cooperation with the International DOI Foundation and tapping work done by the Association of American Publishers (AAP), and the Corporation for National Research Initiatives. According to Tim Ingoldsby, director of business development at the American Institute of Physics (AIP) and a member of the STM board, Wiley and Academic and several others had already participated in the DOI-X pilot project. This project experimented with applying DOI to reference linking and produced the Document Type Definition (DTD) for XML Submission of Metadata and DOIs. Most of the bugs have been worked out.

Ingoldsby also expected the linking process to go very quickly. He said that most publishers have been getting ready for this by tagging all their content since 1997. Now they simply have to gather the metadata and file it with the Central Facility for DOIs. His own institution loaded some 20,000 records over a single weekend. Ingoldsby expects to see a significant body of links in place by the end of February, with the remainder coming into the system by mid-2000.
 

Access Unclear
The press release that announced this program carried a series of quotes from the heads of the different publishing organizations participating in the program. Scanning the quotes, we noticed that the scholarly societies seemed to portray the service entirely as a boon to scientists and researchers, while the commercial publisher statements also pointed out the control publishers could maintain over their products and markets.

Each publisher will set its own access standards, determining what content is available to the researcher following a link (such as access to the abstract or to the full text of an article, by subscription, document delivery, or pay-per-view, etc.). The service is being organized as a not-for-profit entity to safeguard the independence of each participating publisher to set its own access standards and conditions.

When users click on a link in a footnote or endnote within a digitally delivered article, they will go directly to the destination for the next article, or—more specifically—to a publisher permission or access procedure. Ingoldsby indicated that most learned society publishers would provide at least full-text abstracts for free. AIP and some other publishers will also offer pay-per-view options. Users coming from large institutions such as universities with major research libraries may pass through rather seamlessly and experience the full benefits of the integration, since their organizations may already have the licenses in place with publishers.
 

Secondary Publishers—In or Out?
The publishers currently participating and the International DOI Foundation are actively pursuing other scientific and scholarly primary journal publishers to expand the reference-linking program into an industrywide initiative. However, a storm may be brewing over whether they will offer access to secondary publishers to link abstracts and citations with full-text articles.

Baron stated that though secondary publishers (abstracting-and-indexing services) would not appear in the first release, there was "no attempt to exclude them." Elsevier ScienceDirect's Regazzi, former head of Engineering Information, indicated that some secondary services would tap into the new linkages—specifically, those produced by primary publishers already in the program (e.g., Elsevier's Engineering Index/Compendex or Excerpta Medica/Embase)—but only when publishers offered their own abstracting services from their own Web sites, not from third-party search services.

We spoke with two representatives of this group—Jay Trolley at the Institute for Scientific Information, a Thomson company, and Carol Meyer, SilverLinker Product Manager at SilverPlatter Information. Both companies already belong to the International DOI Foundation and have links integrated into existing products—the Web of Science from ISI and the SilverLinker feature on 90 of SilverPlatter's 250 files. However, both agree that access to the new program would improve their companies' products and services.

Meyer points out that SilverLinker already connects to publisher Web sites, including several of those participating in the new program—specifically, Academic Press, AIP, Kluwer Academic, and Springer-Verlag. Meyer stated that they would be happy to participate in the program. Currently they develop connections by getting approval and making arrangements with each individual publisher. One assumes that if they could piggyback on a program run by publishers themselves, this would eliminate a lot of complications and speed the development of better service to users.

Trolley stated, "ISI has been providing cited reference searching for over 40 years. With the introduction of the Web of Science around 2 years ago, we were able to extend the power of citation searching ... ISI is pleased to see publishers acting in concert. We are also pleased to have participated in the process as a charter member of the International DOI Foundation and a member of the DOI-X committee. ISI will expand linking methods to take advantage of the service and provide additional options for customers to reach publisher servers."
 

Wait and See
Difficulties in implementing a successful, comprehensive linking service have mainly stemmed from getting publisher permissions and access to publisher Web sites. This publisher-led development could solve that problem, assuming the publishers offer flexible and workable access. It could also challenge existing services by database aggregators such as SilverPlatter and ISI or even supermarket search services like Dialog and Ovid, by producing a situation in which users and the librarians or information professionals who tend to their data needs end up substituting access to leading publisher titles for access through abstracting services. In turn, this could curtail the visibility of journals from smaller publishers that may rely more on abstracting-and-indexing services for alerting readers to their content.

On the other hand, if scholarly publishers fail to quickly come up with user-friendly access and pricing conditions, the tidal move toward Web-based publishing may swamp them. End users are notoriously quick to form judgments of services. A half-dozen clicks followed by a half-dozen abstract-only, "you are not authorized," or "$40 please" experiences could eliminate whole user populations. Nevertheless, this is a major step toward giving users what they want the way they want it.