Electronic Journals as a Component of the Digital Library

Abstract

Introduction

Digital Library Context

InfoWeb also provides employees with an organized approach to finding information that is openly available on the web, eliminating the frustration of finding thousands of "hits" in response to a web search. The Library has organized frequently requested information into broad subject categories: Computer Support, Government Information, Internet Directories, and Science Resources. Library staff mine the web for sites that contain information of particular interest to NRL researchers. InfoWeb links are then created and annotations are added so that researchers know in advance what they can expect to find at each site that is available through InfoWeb. Every month an automated program checks links to be sure they are active. In addition, the staff continually adds new sites and re-evaluates those that are on the system to be sure they haven't gone out of date. This information is accessible outside NRL and is heavily used by other libraries and end users in government, academia, and the general public.

The InfoWeb resources most in demand by the NRL community are, as would be expected, those that the Library licenses or pays for. These include databases that the Library provides by linking to external web sites, and those databases that the Library mounts locally on its own servers. Examples of licensed databases that reside on remote sites are: the Institution of Electrical Engineers (IEE) INSPEC database covering physics, electrical engineering and electronics, computing and control, and information technology; and OCLC FirstSearch, providing access to more than 40 databases in all subject areas. Locally mounted databases include: Science Citation Index Expanded and the NTIS database of government publications.

Digital Library Implications

One way in which libraries can create content, and ultimately share it on an enterprise-wide basis, is by digitizing institutional publications that are not protected by copyright. The NRL Library has been scanning its technical reports collection since 1989, and has converted about 180,000 reports (nine million pages) to a digital format. Because most of these reports have limitations on dissemination, they are currently available only within the Library to staff and authorized users. As the Library begins to provide services over a secure network, access to all of this content will become more widely available. At this time, the full content of some 6,000 unrestricted reports is openly available through InfoWeb, either through a search of the Library's web-based catalog or through the TORPEDO Ultra digital repository.

A more typical way of providing content for access by library users is to license it. For example, NRL has a license with Elsevier Science permitting it to mount electronic versions of journals on a local server for institutional access. Elsevier was the first publisher to offer electronic journal subscriptions on a commercial basis with a 1996 offering that included 1995 back files. The NRL Library was an early implementer of EES and currently has over 200 Elsevier journals archived in TORPEDO Ultra and available to the user community through InfoWeb.

Cooperative agreements are another approach to building digital content and can help minimize the risk for both publishers and libraries in testing the digital waters. The Technology Transfer Act of 1986 provides the authority for government organizations and other types of organizations to enter into a Cooperative Research and Development Agreement (CRADA). The NRL Library recently completed a CRADA with the American Physical Society (APS) that resulted in the creation of a 10-year digital archive called Physical Review Online Archive (PROLA). Under the terms of this agreement, much of the PROLA content is also mounted locally at NRL. The NRL Library and the American Institute of Physics (AIP) recently signed a similar CRADA agreement, which will expand the AIP's web-based journal offerings and provide NRL users with access to 17 new locally mounted journals going back to 1992.

In addition to locally-mounted content, the NRL Library provides InfoWeb links to over 280 journals that reside on remote sites; examples include the publications of scholarly associations such as the American Chemical Society, as well as journals produced by commercial publishers such as Springer-Verlag.

Digital Library Repository

TORPEDO Ultra allows users to browse the Library's digital holdings in much the same way that users browse a library's physical stacks. For example, a TORPEDO Ultra browse lets users "drill down" to a specific article by selecting a particular journal from an alphabetic or subject list of journals, selecting a volume from the list of available volumes, selecting an issue from the list of issues in that volume, and finally selecting the article from the table of contents for that issue. Users can also search individual journals, groups of journals, reports, or the entire collection. For its search engine, TORPEDO Ultra relies on a commercial off-the-shelf (COTS) search engine called RetrievalWare from Excalibur Technologies. With RetrievalWare, users can perform field searches, say for all articles by a certain author, or they can search the full text of digital documents. RetrievalWare facilitates retrieval by permitting different types of full text searches. A concept search allows users to define terms using an online dictionary. Concept searches can be refined, for example by instructing the system to find "climate" only in its meteorological sense. Concept searches can also be expanded by instructing the system to find terms that are even loosely related to the search term. A pattern search uses a patented algorithm, called Adaptive Pattern Recognition Processing, to find near matches and is useful in compensating for possible OCR errors. A Boolean search option is also available. Search results for concept and pattern searches are presented in a user-defined sort, with the default set for relevance ranking so the "best" hits are displayed first.

Once users identify an article or report they're interested in from the displayed list of results, they can display it in PDF format using the Adobe Acrobat Reader. There are currently two types of PDF files in TORPEDO Ultra: those that have been produced from scanned bit-mapped images (wrapped PDF) and those that are have been generated as part of the publication process (distilled PDF). All PDF files added to TORPEDO Ultra are optimized by the Library for fast retrieval and enhanced with thumbnails of all pages to facilitate document navigation.

Content-Enabling Catalogs, Databases, and Services

The Library has added URLs to its online catalog for some 6,000 technical reports and 10,000 other NRL-authored publications that are available digitally in TORPEDO Ultra. This provides end users performing a web-based catalog search with immediate access to the full content of NRL technical reports and publications. Activating a hyperlink that says "Full Content" launches the Acrobat Reader and displays the PDF image of the first page of the report along with thumbnail page images. A similar process takes catalog users to digital holdings of journals that are in the TORPEDO Ultra collection. In the case of journals, clicking on the hyperlink connects users to the journal in TORPEDO, where they can browse digital holdings or perform a search.

One the most frequently used InfoWeb databases is Science Citation Index Expanded produced by the Institute for Scientific Information (ISI) and mounted at NRL for the use of all employees served by a consortium of four federal science libraries known as the National Research Library Alliance (Stackpole & Atkinson 1998). NRL is currently working with ISI to add hyperlinks to those articles in Science Citation Index Expanded that are stored in TORPEDO Ultra. As new articles are added to TORPEDO Ultra, identifying information and URLs will be sent via FTP to ISI and automatically added to the next weekly database update installed by NRL. A user searching the citation database will see a "Full Text" button for any article that is stored digitally in TORPEDO Ultra. Clicking on the button will display the PDF file. Authorizations built into the database display the button only to users who are permitted to see a particular article, allowing each consortium member to customize its digital holdings and enter into individual license agreements with publishers. In phase two of this project, links to the ISI bibliographic record will be added to TORPEDO Ultra metadata to enable users to link to article references (in the citation database) and display the cited article (in TORPEDO Ultra). In a third phase, links will be added to journals licensed by consortium members for access from publisher web sites.

One of the most popular "digital" services the NRL library has offered over the past several years is called Contents-to-Go. In the beginning, Contents-to-Go had no digital content at all. It combined an automated e-mail alerting service with old-fashioned document delivery (i.e., photocopies sent via interoffice mail).

Here's how Contents-to-Go works. The Library subscribes to electronic tables of contents for all journals in its collection. The tables of contents, which are e-mailed to a library mail server, are automatically redistributed to users who have "subscribed" to the service for specific journals. An InfoWeb interface allows users to select as many journals as they want by clicking a check box. Subscribing requires a user to enter a valid e-mail address, which enables the subscription module to automatically build the mailing lists. No staff intervention is required at any time. Users can add or subtract journals whenever they choose. The e-mail messages that go out to users have customized headers explaining how to request copies of articles, using the e-mail reply function.

An interim step in adding digital content to Contents-to-Go was to add a URL to the header of e-mailed tables of contents for those journals that are stored digitally. This URL takes users into the TORPEDO Ultra system where they can then browse or search for the specific article in which they are interested. Similarly links have been added to Contents-to-Go e-mail messages that take users to journals that are available from publisher web sites. In the very near future the Library will provide users with the ability to link from the e-mailed table of contents to each individual article in TORPEDO Ultra. This will be accomplished by replacing vendor-provided tables of contents for those journals that are part of TORPEDO Ultra with tables of contents automatically created by the TORPEDO Ultra system each time a new journal issue is added. These system-generated e-mailed tables of contents will incorporate URLs for each article providing users with enhanced access to digital journal content.

Hardware Configuration

Network access to restricted InfoWeb services is controlled by IP filtering, using Unix TCP-wrappers and IP address ranges.

Advantages of TORPEDO Ultra

• It permits users to take advantage of NRL's excellent network infrastructure, which provides high speed connectivity and near instantaneous access to locally-mounted content;
• It integrates access to digital content, regardless of whether that content is produced by a commercial publisher, a scholarly association, or a government agency;
• It offers a common interface for browsing and searching digital collections facilitating access and fostering user productivity;
• It creates a local archive for journals and other digital publications, thereby assuring future availability of critical research information;
• It provides an internal infrastructure for reliably and permanently content-enabling other library databases and services; and
• It brings systems enhancements under local control, enabling the Library to provide NRL scientists with a product customized to meet their specialized needs.

Future Development

While NRL anticipates implementing agreements with many more publishers for local journal archiving, in some cases it finds that publishers will agree to provide access to digital journals only from their own web site. In the beginning, the rationale for this resistance appeared to be the fear "losing control" over copyrighted content. However, more recently, NRL is finding that publishers, particularly those that have invested heavily in enhancing their web journal offerings, are increasingly reluctant to allow libraries to load journals locally. Examples of common publisher enhancements are links from references to records in auxiliary databases or digital archives, forward and backward hyperlinking to citing or cited articles, and links to external web sites. The availability of an enhanced hyperlinked web journal appears to cause publishers to view digital versions of the printed publication as inferior products, which they are hesitant to provide to libraries.

TORPEDO Ultra offers a solution to the problem of simultaneously searching or browsing journals on remote web sites along with locally mounted publications. Implementing this solution requires that the journal content, which would remain on the publisher web site, be indexed in TORPEDO Ultra. Users could then browse or search TORPEDO Ultra and click on a hyperlinked title to display the full document in PDF regardless of whether it resides locally or on a remote server. NRL is currently exploring two approaches to provide users with such unified searching of locally mounted and distributed documents. Both approaches, which will probably both be available as publisher options, achieve the same end result; they provide the two types of data required by TORPEDO Ultra: bibliographic records for fielded searching and display, and access to the full text for indexing purposes.

In the first approach, which is currently being tested with the American Physical Society, the publisher has provided sample article content in SGML format. The intent is for NRL to extract bibliographic data from the delivered SGML file and to index the full text of the entire article for searching. An alternative approach, to be tested later this year with the American Institute of Physics (AIP), will allow NRL to use a RetrievalWare Spider to index journal content on the AIP web site; the publisher will provide associated bibliographic data from the AIP SPIN database for use in TORPEDO Ultra.

Conclusion

References

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} [May 17, 1999]

Stackpole, Laurie E. & Atkinson, Roderick D. 1998. The National Research Library Alliance: A Federal Consortium Formed to Provide Inter-Agency Access to Scientific Information. Issues in Science & Technology Librarianship, No. 18, Spring 1998. [Online]. Available: http://www.istl.org/98-spring/article6.html [May 17, 1999]

We welcome your comments about this article.