Access to Periodical Titles in Full Text Databases
by Elaine Day
Database vendors offer a growing number of "aggregated" full text periodicals in subject oriented reference databases. Most libraries omit these periodical titles from local catalogs due to cataloging cost and staffing considerations. Administrators may therefore be chagrined to discover they are paying twice for the same content, or that readers submit interlibrary loan requests for articles available through a local database subscription. Connecting the patron with citation in hand to the full text of an article was the problem addressed in a morning of presentations and discussion by VIVA technology and cataloging specialists on January 7, 2000, at Virginia Commonwealth University.
When a search begins and ends successfully in a full-text database, everyone is happy. The user's need for pertinent full text information is satisfied, and the library receives a fair return on its investment. If, on the other hand, our user starts in a citation database, or is following a tip from a teacher or colleague, her OPAC title search for the cited journal will often fail-unless the library has cataloged individual periodical titles represented in full text databases, or has a separate print or electronic subscription for the cited title.
Catalogers and cataloging administrators have several valid reasons to resist creating or seeking MARC records for aggregated journals in Dow Jones, the Gale InfoTrac databases, Lexis-Nexis, and others:
- The number of titles involved is large; serials cataloging and maintenance is a costly and meticulous process.
- The selection of titles offered in a database may be volatile, with frequent, unannounced additions or deletions.
- Full text coverage is not always complete. Some articles, advertisements, features, or graphics are omitted from the electronic version for copyright or cost reasons.
- Database vendors, with the exception of EBSCO, which has contributed to an ALA ALCTS demonstration project, do not provide MARC records for their holdings.
- ISSNs are sometimes omitted from vendor title lists.
- The single record approach used by many libraries to describe multiple versions of the same periodical title (print, microform, electronic) does not lend itself to rapid changes in licensing. Parts of the record must be edited when a subscription is added or removed.
- Most importantly, in contrast to traditional OPAC functionality, the user is unable to go directly from the catalog to the information he seeks, but must perform a secondary search in the full-text database to locate a specific title, volume, and issue.
Periodical databases generally offer a journal title search. Some vendors publish their title lists on the Web. Merging vendor title and holdings information into a single interface with other periodical lists and catalog records was the specific challenge addressed by the VIVA forum.
In her opening remarks, Eileen Hitchingham of Virginia Tech mused that in a perfect information gathering world, users will have access to a mega database for all current periodical literature. Or, perhaps databases vendors will link their holdings so that citations in one interface point directly to full-text in another product for authenticated subscribers. To make retrieval even more convenient, users will click on bibliographic references leading directly to other full-text articles. Although falling short of this vision, our present environment offers several low-cost, interim solutions for title access:
- a "flat file" periodical title search including aggregated and cataloged titles (Radford University),
- a simple database external to the public catalog (Old Dominion University, Virginia Commonwealth University), and
- brief MARC records created for the public catalog from vendor data, with "hooks" for maintenance (Virginia Commonwealth University).
Doris Hendrickson of Radford University illustrated the flat file periodical search. Selected fields from periodical records in Radford's INNOPAC system are listed to a text file, then processed with a Perl script adapted from one developed by Ed Summers for Old Dominion University. The resulting file includes title main entry and alternate titles, providing nearly the same level of title access as the library catalog. Hendrickson downloads vendor title lists from the web and processes them with additional scripts that address the peculiarities of each vendor's web format. She combines the resulting files into a single file of aggregated titles. Another script merges the catalog and vendor data, concatenating identical titles into a single entry for the final flat file. The data file is then installed on the public web and searched with Perl and CGI. Radford's Periodical Search has proven hugely popular with library users.
Old Dominion University's Ed Summers illustrated an analogous approach to periodical title access. With over 20,000 cataloged periodicals, ODU found the flat file approach too slow for searching. By scripting the fields listed from their INNOPAC and vendor title data to create records and fields, Summers and his colleagues were able to convert the information to a database structure that is searchable with the Apache web server and MySQL-a freeware version of the SQL search software. As of January 2000, the periodicals database at ODU, with 24,894 catalog and aggregator records, had been in operation for more than a year. It had been searched a total of 53,639 times. In November of 1999 alone, it was searched an average of 315 times a day.
Jimmy Ghaphery from Virginia Commonwealth University Library described a similar database, including aggregated electronic titles only. VCU's Online Journals Search is created with MS Access and delivered on the Web using Allaire's Cold Fusion software. The Radford and VCU periodicals search both feature pop-up database search tips developed at VCU with JavaScript. As VCU enters these titles in the public catalog, a special concern is the absence of ISSN's on files from Dow Jones Interactive. VCU staff have identified 1549 out of 4500 ISSN's desirable for MARC cataloging.
Barbara Anderson, also of Virginia Commonwealth University, outlined VCU's method for creating batches of MARC records from vendor title lists. Using a web template designed by Ghaphery, staff generate records using LC MARCMaker software. To facilitate catalog maintenance, each record includes a computer-assigned identification number for the database vendor. After customizing the records for the local OPAC, MARCMaker converts them to standard MARC transmission format for import to the library catalog. Because VCU chose to maintain the traditional "separate record" approach, they are not hampered by maintenance issues associated with single records for multiple versions. New subscriptions can be imported and cancelled subscriptions globally deleted without the need to edit fields in existing MARC records.
Anderson's model is a low cost, low maintenance approach to MARC cataloging that integrates print and electronic subscriptions with aggregated titles. The user is not required to search different indexes for different material formats or try to guess which format will carry the content he is seeking.
In the last presentation of the morning, Elaine Day of James Madison University and Allison Sleeman of the University of Virginia illustrated examples from other institutions and cataloging examples from UVa's VIRGO system. Initiatives of special interest are:
- the Buddy Project at the University of Melbourne, which dynamically delivers all electronic resources offered by the university libraries via a massive web database with cataloging, analytics, and authority control, and
- Yale University's Jake project, an open source database that focuses on cooperation with vendors and other libraries to provide dynamic bibliographic access to periodicals and specific articles in commercial databases. The Jake database and software for building a local version of Jake are freely available to all, although access to the aggregator content is not.
MARC cataloging is also a popular option. Day reported that a survey by The ALA ALCTS Program for Cooperative Cataloging, Standing Committee on Automation, conducted before ALA Midwinter 1999, found that 71% of the 62 responding libraries want records in the OPAC to represent full-text journals available from aggregators, and 75% are interested in purchasing record sets. The University of Tennessee is creating core MARC records from Dow Jones and Proquest data, using Perl scripts and LC MARCMaker software. The University of Virginia focuses on full MARC cataloging of electronic resources. Allison Sleeman illustrated VIRGO records for periodicals in the Gale Infotrac databases.
How can VIVA facilitate title access to periodicals in aggregator database? A wrap-up discussion followed the presentations, with general interest in several possible scenarios and strategies:
- The VIVA central office downloads the data on a monthly basis from vendor websites and posts it, unprocessed, to the VIVA web.
- VIVA uses its influence to lobby vendors for standard data formats, including ISSN's.
- Downloads are coordinated at one VIVA campus, which performs some preliminary processing.
- VIVA works with Jake developers to find out what the two consortia can do to maximize the value and stability of (a) downloads and (b) links to titles and issues.
- VIVA libraries are encouraged to develop and share utilities for managing journal access, such as the VCU web form for creating MARC records from aggregator data, scripts for processing vendor data, and scripts for managing output from various integrated library systems.
- Technology and cataloging expertise are shared by larger VIVA campuses with smaller campuses.
- The VIVA Task Force on Cataloging and Intellectual Access explores options for identifying missing ISSN's on vendor title lists.
- The Task Force continues to monitor serials cataloging rules and guidelines pending from AACR and CONSER.
- Database vendors are encouraged to supply MARC records for aggregated titles.
- VIVA and database vendors explore XML as a MARC or metatdata transmission format
- We keep it simple! Gaps in full text access represent a temporary problem that will ultimately be resolved with database integration and an increasing volume of well-indexed full text sources.
Given cooperation among Virginia libraries and database vendors, the January 2000 forum will be a bridge to Eileen Hitchingham's vision: linking from citation database to full text database, from database to local holdings, and from bibliographical references to full text articles. The forum was also a healthy exercise in cooperative bibliography, technology sharing, and brainstorming!
Reference
VIVA, the Virtual Library of Virginia. Forum on Access to Electronic Journals from Aggregator Sources http://www.viva.lib.va.us/viva/tech/cat/journal_access/
The Forum homepage includes links to presentations and resources.
