The Google Scholar project (http://scholar.google.com), which launched in November 2004 (http://newsbreaks.infotoday.com/nbreader.asp?ArticleID=16324), has responded to the complaints of many academic and research librarians by expanding its usefulness for campus-based users. Its new institutional access feature links Google Scholar users to electronic versions—and even print versions—of journals accessible through library collections. Any library using OpenURLs and meeting Google Scholar's conditions can join the program. Authorization of "appropriate copy" to individual library patrons, "on-campus or off," remains the library's electronic responsibility. Unlike many commercial information services, Google offers the institutional link resolving at its usual attractive rate—free. Within days of the announcement, a reported 150 libraries had joined.
Approximately 30 libraries and several major library software vendors offering link resolvers (e.g., SFX from Ex Libris, Article Linker from Serials Solutions, and 1Cate from Openly Informatics) have worked since the beginning of this year on a pilot project to develop and test this important new feature. Though its scholarly focus clearly targets academic and research libraries, any and all libraries can participate. "We'd like to see them all," welcomed Anurag Achaya, Google Scholar's project manager. Some public libraries have joined already. Although U.S. libraries dominate, participants stretch from Iceland to Japan, Beirut to Tel-Aviv.
The original announcement of the service promised to provide a list of libraries participating in the program at this site. However, Google changed it to a search by name as the list of libraries lengthened so rapidly, according to Achaya. (For those still interested in seeing a list of participants, Achaya suggests some creative searching—e.g., "university," "library," etc., to pull up partial but lengthy lists.) Google Scholar had already offered some OpenURL linking, but only what met DOI (digital object identifiers) and PMID (PubMed unique identifier) standards. Now it has expanded its linking to a broader range of OpenURL connections.
Librarians interested in linking their holdings through Google Scholar should start by going to the "Support for Libraries" page (http://scholar.google.com/scholar/libraries.html). The page also describes the "Library Search" function in Google Scholar that connects to scholarly texts identified from OCLC's Open WorldCat collection. Libraries that already have link resolvers in place may merely need to reconfigure their system. Libraries using "home grown" link resolvers have to do a little more work, but Google will accommodate them. Librarians can also choose different labels for material available as full text online and as print holdings.
Google insists that libraries supply it with their electronic holdings information—journal titles and subscription dates—directly. Google promises not to share this information with any third parties, nor to share any usage statistics, even "aggregate usage based on institutional characteristics or profiles." However, Achaya admitted that libraries themselves might set up systems to track usage of Google Scholar for reaching their collections.
Users can set their own "preferences" by identifying one or more libraries to which they have access (http://scholar.google.com/scholar_preferences). Once set, the preferences will generate highlighted links to online journals that the libraries have licensed. Google needs IP address ranges to automatically connect the service to patrons using library networked equipment, eliminating the need for users to individually re-configure their preferences. Connections to full-text online articles will have special emphasis, but even journals available only in print on library shelves may have links in place. According to Achaya: "If we know the library has an online version of the full text, the link is highlighted and moved up. We want to give the user an indication of access."
The burden of verifying institutional access rights remains one for libraries to solve through passwords, campus computer restrictions, or configuring browsers to a library proxy. Achaya described Google Scholar's role as one of "discovery, not authentication." However, he also indicated that if a patron searcher had more than one affiliation, e.g., multiple campuses, it would be easy to configure.
Libraries that join the program can choose to withdraw holdings information, fully or partially, a request that Google promises to process within 30 days. Since Google Scholar retrieves its content from open Web harvesting and direct deals with content providers, Achaya points out that the results will not change for the user; the library access indicators would just go away.
Reactions to the new institutional access feature in Google Scholar were generally enthusiastic. Gary Price's ResourceShelf coverage of the announcement (http://www.resourceshelf.com/2005/05/be-it-resolved-that-google-scholar-is.html) stated:
Google deserves kudos for opening up this service to all libraries. However, even as the Google Scholar database continues to grow, we still don't know precisely when or how often it's updated, the lag time (if any) for material to get into the database, and other important facts like what will or will not be included in the database. It would also be great if Google could provide a list of sources to which they are providing access.
Jill Grogg, reference librarian at the University of Alabama libraries and author of several articles on linking, wondered: "How will this affect the mass implementation of federated search engines at libraries? Is there room for Google Scholar and federated search products? And the question I always have—what will this do to subject-specific databases? Will they be able to survive in the wake of Google Scholar?" In fact, Grogg reported talking to a colleague who was considering upgrading to a better federated search product but now wondered whether "if we just sit on it for a year or two, Google Scholar will possibly replace federated searching." Grogg speculated, "Maybe there is a place for both."
Achaya does not consider Google Scholar to be a danger to federated searching. "We're not competing with them. We're searching articles accessible to us. Federated searching is finding what the library has access to."
John McDonald, acquisitions librarian at the California Institute of Technology's library, a member of the pilot project, said that any library already using a link resolver would probably find connecting to Google Scholar easy. McDonald pointed out that a study had shown that the restricted (DOI/PMID only) linking previously offered in Google Scholar only reached a third of its citations. Under the new institutional access, almost all of the journal titles would be reached. However, he did point out that other material in Google Scholar, e.g., conference proceedings or theses and dissertations, would still not link to library collections.
Achaya also confirmed that the full text of books remained largely unavailable through Google Scholar. Not only does it have no plans to integrate Google Print holdings, even though millions of books will soon start coming online through the Google Print library project (Google-brary?), but even holdings of primary material, e.g., the massive public domain content in Project Gutenberg, are only available through the main Google search. McDonald considered the main Google service as still offering the bulk of Google Scholar's content plus, in many cases, more material.
In discussing this new development, a difference in viewpoints emerged between Google Scholar's Achaya and some of the librarians with whom I spoke (including the librarian writing this NewsBreak). Achaya takes a scholar's view of the product. He focuses a lot of his software development efforts on the interesting challenge of "versioning," i.e., grouping different versions of the same work. Nor does Achaya define a work in the same way that librarians would, namely as a document or set of documents; he seems to define it more as a coherent scholarly effort. For example, while his versioning efforts encompass pre-prints, post-prints, self-archived journal article texts, and publisher-born articles, it also extends to conference presentations and technical reports. And, while librarians might focus on issues of access, Achaya thinks of that as secondary to discovery, "versioning," and ranking.
McDonald's complaints about Google Scholar, even with its new linking expansion, focused on access issues. For example, he wanted a broader use of OpenURLs, one willing to accept partial information and to offer the outreach options that libraries make through interlibrary loan, document delivery options, sister library connections, etc. He even grumbled that they should re-title the service Google Scholar Full-Text. McDonald recognized Google's generosity in offering all its services for free, but he also saw that this policy left librarians with little or no leverage, as with traditional paid vendors. However, he asserted very strongly that Google Scholar would not succeed without support from academic librarians. McDonald wondered whether the problems in Google Scholar didn't stem from a lack of understanding within Google of the library perspective. He "had the impression that Google and libraries just need to explain our positions better and we would meet each other halfway. Google Scholar has a great chance for success and libraries should be involved, but each of us should do our own thing."