The National Research Library Alliance Issues in Science and Technology Librarianship Spring 1998 DOI:10.5062/F4MK69W6 URLs in this document have been updated. Links enclosed in {curly brackets} have been changed. If a replacement link was located, the new URL was added and the link is active; if a new site could not be identified, the broken link was removed. The National Research Library Alliance: A Federal Consortium Formed to Provide Inter-Agency Access to Scientific Information Laurie E. Stackpole Roderick D. Atkinson Ruth H. Hooker Research Library Naval Research Laboratory Washington, DC 20375-5334 laurie.stackpole@nrl.navy.mil rod.atkinson@nrl.navy.mil Abstract This article outlines the circumstances that led to the formation in the Spring of 1997 of the National Research Library Alliance (NRLA), a consortium of federal libraries. It discusses some of the issues that face organizations when they attempt to act collectively and the need for vendors and libraries to work cooperatively to solve the inherent problems associated with a consortium model. The NRLA's first joint procurement, a license to mount one of the Institute of Scientific Information's "Web of Science" databases, was finalized in July 1997. This procurement, which provides some 5,000 researchers at four institutions with web access to Science Citation Index Expanded, is used throughout the article to illustrate some typical problems that arise when diverse institutions attempt to work in unison, as well as to illustrate the benefits that result from forming partnerships to license critical information resources. Introduction To provide their research communities with a common core of scientific information, four federal libraries have formed a consortium called the National Research Library Alliance (NRLA). NRLA members are the NASA Goddard Space Flight Center Library, the Office of Information Services of the National Institute of Standards and Technology, the National Science Foundation Library, and the Ruth H. Hooker Research Library of the Naval Research Laboratory. The first product licensed to NRLA was the "Web of Science" Science Citation Index Expanded, produced by the Institute for Scientific Information (ISI) (ISI Press Release 1997). Science Citation Index Expanded provides a web interface for searching over 5,000 major scientific and technical journals. Science Citation Index Expanded, covering the years 1985 to the present, is mounted at the Naval Research Laboratory, providing over 5,000 researchers at the four participating institutions with desktop access. Background The Ruth H. Hooker Research Library at the Naval Research Laboratory was an early adopter of technologies to provide its user community with desktop access to information. In 1992, it implemented the InfoNet campus-wide information utility, a character-based system providing NRL employees with telnet access, from work or home, to CD-ROM products, Laboratory databases, and selected Internet resources (Atkinson et al. 1994; Atkinson & Stackpole 1995). One of the first CD-ROM products to be networked via the InfoNet was Science Citation Index, which proved to be one of the InfoNet's most popular resources, generally trailing only the Library's online catalog in number of monthly accesses. While InfoNet services met many of the needs of NRL researchers for information, in early 1996 the Library, with the guidance of a study team convened by the NRL Director, determined that CD-ROM networking had largely served its purpose (InfoVision/2000 Study Team 1996). In the Spring of 1996 the decision was therefore made to migrate InfoNet resources to the web-based InfoWeb Information System and Gateway ({http://infoweb.nrl.navy.mil}) as quickly as practicable, but certainly within two years. The product that appeared to be most problematic in making the transition from InfoNet to InfoWeb within the desired time frame was Science Citation Index. As this was one of the InfoNet's most heavily used databases, the Library was eager to find a way to provide web access to this important research tool. At that time, the Institute for Scientific Information had not released its plans for a web offering. The alternative of licensing the information for local access and undertaking the custom development of an NRL web product appeared out-of-reach for a relatively small library, both in terms of its cost and the level of effort required for custom programming. Data, Data, Who's Got the Data? The authors visited the Library of the Los Alamos National Laboratory (LANL) ({http://library.lanl.gov/}) in the Fall of 1996 to exchange information about the respective digital library initiatives of the two organizations. During that visit NRL learned that the LANL Library, under the direction of Library Director Richard E. Luce, had developed customized software for web access to Science Citation Index, called SciSearch at LANL, using data licensed from ISI, delivered in tape format, mounted on a local server, and searchable using the Verity search engine. Furthermore LANL, with the agreement of ISI, was serving as a Science Citation Index host site for members of the New Mexico (Library) Alliance, which had independently contracted with ISI for access to a subset of the data mounted at LANL. After successfully exploring the feasibility of contracting with LANL for NRL access to SciSearch at LANL, the Library contacted ISI to discuss data licensing arrangements and pricing. It was immediately clear that NRL acting on its own would be hard pressed to cover the significant costs associated with licensing the data required even when credits were applied for prior investments and rights granted under the CD-ROM license agreement. Fortunately the NRL Library was not the only federal library with a keen interest in providing Science Citation Index to its user community. Three other federal science libraries had long-standing network licenses for the Science Citation Index CD-ROM product and were equally eager to move more fully into a web environment. The other three libraries served clientele at the NASA Goddard Space Flight Center Library in Greenbelt, MD ({http://www.nasa.gov/centers/goddard/home/index.html}); the National Institute of Standards and Technology in Gaithersburg, MD (http://www.nist.gov/), and the National Science Foundation Library in Arlington, VA (http://www.nsf.gov/). The NIST Office of Information Services was actively developing its NIST Virtual Library program (https://www.nist.gov/nist-research-library); the NASA Goddard Library was offering a range of services through its Centerwide Virtual Library ({http://library.gsfc.nasa.gov/}) (Dixon 1997); and the NSF Library was providing its staff with databases over an Intranet. A Collective License -- Necessary and Sufficient Conditions Two sets of conditions proved critical to the ability of the four interested libraries to act together. The first was sufficient commitment on the part of the four library managers to "hang in there" when the going got tough and to seek creative solutions to the barriers that at times threatened to sink the concept of an interagency alliance. The other condition was flexibility and cooperation on the part of the vendor, who consistently came up with innovative solutions to help the libraries work within their financial, legal, and management constraints, which often differed significantly from institution to institution. Library Commitment From the start there was a great deal of cohesiveness and shared sense of direction among the four libraries that eventually formed the NRLA consortium. All four libraries had a long-standing history of cooperation and shared a similar philosophy of information delivery and user empowerment. Furthermore, all were at a comparable stage of technological development and were able to capitalize on sophisticated organizational network infrastructures. Last, but not least, they all served scientific user communities that had many common characteristics and research interests. While all prospective NRLA members endorsed the idea of forming a federal library consortium to obtain more advantageous pricing for a variety of information products (such as databases and electronic journals), without the shared enthusiasm for a particular product, NRLA probably would not have become a reality. NRL's interest in rapid deployment of web access to Science Citation Index served as an added incentive to keep the project on target and on track. While a keen interest on the part of the libraries in adopting a consortium model and applying it to the procurement of a critical information resource was a prerequisite, other agency offices had roles to play. Without the combined efforts of various staff, particularly legal staff who reviewed iterative versions of the ISI Database Lease Agreement and the NRLA Memorandum of Understanding, a collective purchase could not have occurred. The fact that all four institutions were federal agencies would seem on the surface to be a common denominator that would simplify the procurement process. However, in reality there were almost as many differences among the various agency approaches to contracting, licensing, purchasing, and interagency collaboration as there were similarities. One of the most basic differences was the fact that NRL as a Department of Defense agency operated under a different set of procurement guidelines than the other three agencies. For NRL, the Defense Federal Acquisition Regulation Supplement (DFARS) governed the terms and conditions of any licenses that the agency accepted. For the other three agencies, the Federal Acquisition Regulation (FAR) was operative. While in most cases the FAR and DFARS are in sync, in at least one key area, e.g., rights to the use of technical data, there is a conflict that could become important were there ever to be a dispute with regard to use of the ISI database. Ultimately, this was resolved by all agencies agreeing to apply the more restrictive clause, which in this case was the DFARS. Vendor Flexibility With the New Mexico Alliance as its prototype, NRL began to discuss with ISI the concept of licensing the Science Citation Index database in concert with its three prospective partners. ISI agreed to quote a single price to the NRLA consortium, leaving it to the member libraries to determine how they would allocate the costs among themselves. Credits were applied to the total cost for all partners' prior rights in the Science Citation Index data resulting from purchases of CD-ROM products and tape licenses. In many consortium models, the consortium acts as an independent entity and enters into a contractual arrangement with a vendor. While this model could potentially be implemented in a federal government context, it was not immediately clear to the NRLA participants how such a centralized procurement could be effected in a timely manner and in a way that would overcome various concerns, including legislative and administrative requirements. ISI recognized the constraints under which NRLA members were operating and, instead of insisting on a single consortium payment, ISI worked with agencies to develop individual payment models that would comply with each member's procurement requirements. The pricing model that NRLA adopted was an even distribution of costs with each participant covering one quarter of the total cost. The Web of Science The announcement by ISI in 1997 of its release of a commercial web-based product that could be mounted locally, provided an alternative to SciSearch at LANL and was the solution that the NRLA consortium eventually adopted. While ISI requested that all members of the consortium agree to a single method of access, the company did not stipulate whether that would be through Los Alamos or by mounting the data locally at one of the NRLA sites (the only option offered for Web of Science access at that time). Decision Several factors entered into the decision to mount the data as a local Web of Science offering rather than to contract for access to SciSearch at LANL. In the end the decision came down to two major considerations: ISI corporate capabilities and the desire for local control. There was the tacit assumption on the part of NRLA members that ISI had a unique and unparalleled interest in providing the best possible showcase for its own data. It was therefore believed that going with the ISI-supported product, with a regular cycle of software upgrades and product enhancements, would afford maximum protection for NRLA's investment. In addition, ISI demonstrated a corporate commitment and ability to provide technical support. Also important to the NRLA members was ISI's ability to support end-user outreach and training with professional instruction and materials. Other issues that entered into the decision to license the Web of Science version of Science Citation Index Expanded were concerns over system performance and response time with a very large user base and a desire for greater local control to customize future enhancements to NRLA specifications. Implementation The NRLA Science Citation Index Expanded database went live in July 1997 with a test database covering 5 years worth of data ({http://infoweb.nrl.navy.mil/NRLA/}). In October 1997 the database was migrated to a new NRL server and coverage was extended back to 1985, providing users with a 13-year data file (Oxley 1998). The database is available for desktop access by more than 5,000 researchers at the four participating institutions, enabling them to locate publications through a subject or author search, or to find articles that reference relevant authors or institutions. The references within indexed articles can be used to find other relevant materials or to establish relationships among articles. Hyperlinks enable users to navigate between references and cited articles when those articles have been indexed in the database. Unlike its predecessor products, the web version permits searches for articles citing a particular author even if the author is not the first-listed, or senior, author. Science Citation Index Expanded runs on a Sun Enterprise 4000 under the Sun Solaris operating system. The database, which requires just under 50 GB of online storage, is accommodated by a Sun Storage Array RAID level 5 system. Weekly updates to the database are downloaded from ISI's corporate server via Internet FTP (File Transfer Protocol). Network access to Science Citation Index Expanded is controlled by IP filtering, using Unix TCP-wrappers and IP address ranges. Since all members of the consortium subscribe to the same 13-year subset of the database, additional access controls are unnecessary. However, the ISI Web of Science software includes an administration tool that can restrict access to portions of the database using appropriate IP address ranges. This tool may be helpful to NRLA in the future should some members of the consortium opt to purchase additional years. For all members of the consortium, this access control arrangement has proved ideal; both administrative and technical overhead are low and access has been seamless for the vast majority of users. In addition, while no Internet security system is infallible, these access controls are sufficient to stop all of the casual, and even many sophisticated, unauthorized users. Usage statistics for March 1998, the most recent full month for which data are available, show over 3,000 accesses to Science Citation Index Expanded by NRLA end users. Roughly one-third of the use is by NRL users, one-third by NIST users, and the remaining one-third is divided between the other two partners, with NASA responsible for two-thirds of that usage and NSF one-third. At NRL statistics show that the web version of Science Citation Index is accessed roughly three times more often than was the networked CD-ROM product previously available through InfoNet. Formalizing the Consortium As its operating charter, NRLA members drafted a Memorandum of Understanding (MOU) setting forth the consortium's purpose and principles. The purpose of the alliance is to "improv[e] access to a common core of scientific information required by their respective user communities." The principles, as developed by consortium members, were both few and simple. They briefly outlined how decisions on consortium procurements would be made, how product and operational costs would generally be allocated, and common expectations that effort would be equitably distributed and that acting as a group would ultimately save money. The MOU was drafted in July 1997, and subsequently passed through the required chain of command in each institution. As various organizations reviewed the MOU, each began to add its own unique requirements. New concepts were incorporated and the list of principles became longer and more complex, with specific references to Executive Orders and to the U.S. Code. Each major change needed to be reviewed by offices in other institutions that had already approved a prior version of the MOU, making many iterations necessary. At this time, the NRLA MOU is still under review. Conclusion The savings realized by each consortium participant in licensing the ISI Science Citation Index Expanded, 1985-1998, were significant. In addition, the NRLA procurement provides member libraries with greater "bang for the buck" through expanded database coverage, improved search capabilities, and a larger user base. Annual savings in subscription costs in future years will also be realized under the consortium license, with NRLA members paying little more than they had for the networked CD-ROM product. The cost of maintaining and administering the database and the associated Web server and RAID storage is shared by all consortium members. For three of the participating libraries, the time involved in providing their user communities with access to a critical information resource has been virtually eliminated. For NRL, where the database is physically mounted, providing web access to the Science Citation Index Expanded for the NRLA consortium has been found to require the ongoing services of a skilled UNIX administrator. At NRL, web access to Science Citation Index Expanded has successfully replaced the Science Citation Index CD-ROM previously available through the InfoNet. With the successful migration of the full range of scientific databases provided by InfoNet to InfoWeb, the NRL Library was able to meet its target date for discontinuing its character-based service. References Atkinson, Roderick D., Stackpole, Laurie E. & Yokley, John. 1995. Developing the Scientific-Technical Digital Library at a National Laboratory. In: Digital Libraries: Current Issues, Digital Libraries Workshop DL '94, Newark, NJ, USA, May 1994, Selected Papers (ed. by Nabil R. Adam, Bharat K. Bhargava, and Yelena Yesha), pp. 265-279. Springer-Verlag, Berlin. [Online]. Available: {http://infoweb.nrl.navy.mil/NRL_publications/digital_library_94.html} [April 21, 1998]. Atkinson, Roderick D. & Stackpole, Laurie E. 1995. TORPEDO: Networked Access to Full-Text and Page-Image Representations of Physics Journals and Technical Reports. The Public-Access Computer Systems Review 6(3). [Online]. Available: {http://epress.lib.uh.edu/pr/v6/n3/atki6n3.html} [April 21, 1998]. Dixon, Robin M. 1997. Goddard Library's Web Page. In: The Library Web (ed. By Julie M. Still), pp. 79-88. Information Today, Inc., Medford, NJ. InfoVision Study Team. 1996. Report of Findings and Recommendations Submitted to the Director of Research, Naval Research Laboratory, September 1996. [Online] Available: {http://infoweb.nrl.navy.mil/NRL_publications/digital_library_94.html} [April 21, 1998]. ISI Press Release. 1997. "NRLA Consortia Purchases the Science Citation Index Expanded(TM)." Philadelphia, PA, USA, October 24, 1997. [Online] Available: {http://www.isinet.com/whatsnew/announce/pr971024.html} [April 21, 1998]. Oxley, Harriet. 1998. ISI Spins a Web of Science. Database 21(2), 37-40. [Online]. Available: {http://www.onlineinc.com/database/awards/award3.html} [April 21, 1998].