Electronic resources are becoming increasingly important to libraries and library patrons. While resources such as stand-alone CD-ROMs or diskettes are still physical possessions of the library, network-based resources, whether local or remote, do not fit this model of traditional library materials. Electronic resources challenge traditional library modes of operation, from collection development through bibliographic control and public services. This working group was formed to make recommendations about whether network-based resources should be cataloged in HOLLIS, and if so, which resources should be included and how they should be treated.
A fundamental fact underlying the group's discussions was the incredible speed with which network-based resources and retrieval tools are evolving. The task force agreed that its recommendations had to focus on the near term, to be very flexible, to allow as much benefit as possible without a huge investment, and to recognize the experimental nature of trying to bring electronic materials under bibliographic control.
Bibliographic records perform several functions:
For network-based resources, some of these functions can be performed by various Internet retrieval mechanisms, but none of the mechanisms yet works for all kinds of network-based resources, let alone brings electronic resources together with print and other traditional resources. For this reason, the group agreed that there is value in cataloging selected network-based resources in the library union catalog. At the same time, however, there was a recognition that we cannot "catalog the net", and that in the future no one system of databases, such as HOLLIS, can be considered to describe the library's offerings comprehensively. A suite of library applications including HOLLIS and HOLLIS Plus will be available to notify patrons of library resources.
RECOMMENDATION 1: Policies, guidelines, and technical changes should be put in place to support cataloging of network-based resources in HOLLIS.
Identification is one of the more problematic aspects of cataloging electronic resources. In many cases it is difficult to describe a given resource unambiguously (e.g., what makes one university's Academic Index database different from another's?). Unlike a print serial, where each new issue triggers some level of verification, there is usually nothing to notify a library when a resource is modified and the bibliographic information should be changed. Also, in an extension of the multiple versions debate which has raged in cataloging circles for years, there is no agreement on what constitutes an intellectually equivalent version or a separate edition of an electronic resource. The group agreed that, given the lack of control the library has over network-based resources, the same level of identification supplied for traditional resources may not be practical or necessary for all electronic resources.
RECOMMENDATION 2: Where appropriate and applicable, network-based resources should be cataloged on the same bibliographic records as their print equivalents. Where appropriate and applicable, multiple manifestations of a resource (e.g., different archives of one electronic journal, different formats of a single document) should be cataloged on the same record.
The types of resources available on the network range from documents to databases to services, from materials which are very analogous to traditional library acquisitions to online services and sites which have no print, film, or vinyl equivalents. The group discovered that perceptions of what resources should be cataloged varied from library to library, and agreed that individual Harvard libraries have the option of deciding whether, which, and how many networked resources to catalog. We agreed that the selection process should identify particularly valuable resources and thus help patrons avoid the overload of Internet-wide searches. We identified a set of criteria which staff can use to decide whether to catalog a particular resource. Also, we acknowledged that putting a resource in HOLLIS will involve not only the initial investment in cataloging, but also an ongoing commitment to the maintenance of that catalog record to insure its accuracy. Further, reference staff throughout the university will have to instruct users in the interpretation of the records and address questions about the resources described. For these reasons libraries should weigh the costs of cataloging the resource against the benefits to be gained by collocating it with traditional materials.
RECOMMENDATION 3: The criteria below (which include traditional selection criteria) should be considered when deciding whether an investment in cataloging should be made. Units choosing to catalog networked resources should weigh these factors, together with the commitment in staff time, when deciding whether a given resource should receive cataloging:
For example, resources most suitable for cataloging might be those that Harvard has paid for and/or owns and is therefore committed to maintaining, those that another university is maintaining, and government resources which have historically been published in print and which now are published only electronically.
The group discussed how to decide whether individual records should be created where several documents or databases are available in a service. This case is somewhat analogous to microform sets, which have historically received a range of treatments, from a single record for a set to individual records for each filmed title, depending on the library's perception of the value of the material. It is true that selective cataloging of individual items within a resource (for example, the journal titles available through Uncover) might mislead a researcher to assume that all individual titles are cataloged. Judicious selection of titles, however, might call attention to important, but otherwise invisible titles. We believe that there cannot be a hard and fast rule that all individual titles or none be cataloged and would advise units to be aware of the implications of selective cataloging.
The group considers it important to use a HOLLIS LOC which is not library-specific for those networked resources which are widely available, to avoid leading users to think that they must come to a specific library to use the resource. Specific library LOCs would only be used for those resources which can only be accessed through local area networks, i.e., those resources which do require the user to be at a specific device or devices.
RECOMMENDATION 4: Implement a new LOC code, "net", which would display in the Harvard locations area as "LOCATION: Networked Resource:" This should be used when cataloging resources available through HOLLIS Plus or resources widely available on the Internet. Libraries should use their own LOC codes when cataloging resources which must be used in the library (or at some limited set of physical devices). In this case, the sub-location "$c Networked Resource" should be used. The implication of this recommendation is that only one unit need catalog any given resource, and that the rest of the community will share that information, and can enrich it as appropriate.
HOLLIS must be able to handle both textual and structured access information in order to provide the greatest flexibility for staff and the greatest utility for users. While it may not be anticipated that HOLLIS will support a link from the catalog record to the resource, the group agreed that some method of providing this functionality is not far in the future and that the Harvard libraries should position themselves to take advantage of it.
RECOMMENDATION 5: The 856 field should be implemented in the HOLLIS holdings record. For non-serial records loaded into HOLLIS, the presence of an 856 field should trigger the building of a linked holdings record. The field should display in the LOCATIONS area of the bibliographic display with the literal "To access:" All subfields should be valid, but only the following subfields should display, in this order:
|$u||Uniform Resource Locator||URL:|
|$s||File size||File size:|
LOCATION: Houghton: bMS Ger 146-146.3 Access may be restricted. Details at the repository. To access: Finding aid URL:gopher://elmer.harvard.edu/11/.links/archives/.indrep/.hough/.houghfind/.houghfeinTechnical services:
Feininger, Lyonel, 1871-1956. Papers, 1883-1960 (inclusive). LOC/1:9c: $i hou $b bMS Ger 146-146.3 $k nethou HU CREATED 07/09/86 UPDATED NONE FORMAT: H RECSTAT n RECTYP: x LVL: 1 STAT: ` METH: ` RET: ```` COM: ` COP: `` CAN: ```` LANG: ``` 506/1: : $a Access may be restricted. Details at the repository. 541/2: : $3 Papers. $a Feininger Family, 1979, and deposit, 1963, from Busch-Reisinger Museum. $c Gift and deposit. 583/3: : $3 Papers. $a Survey/esp $c 06/12/86 $k dw 856/1: : $3 Finding aid $u gopher://elmer.harvard.edu/11/.links/archives/.indrep/.hough/.houghfind/.houghfein
LOCATION: Networked Resource: To access: URL:ftp://ds.internic.net/rfc/rfc1437.txt File is in ASCII text. File size: 26K To access: URL:ftp://ds.internic.net/rfc/rfc1437.ps.gz File is in Postscript format. Compression: Must be decompressed using GUNZIP File size: 21KTechnical services:
TCP and UDP with bigger addresses (TUBA) : $b a simple proposal for internet addressing and routing. LOC/1: c: $i net $k netcab HU CREATED 01/23/95 UPDATED NONE FORMAT: H RECSTAT n RECTYP: x LVL: 1 STAT: ` METH: ` RET: ```` COM: ` COP: `` CAN: ```` LANG: ``` 856/1: : $u ftp://ds.internic.net/rfc/rfc1437.txt $z File is in ASCII text $s 26K 856/2: : $u ftp://ds.internic.net/rfc/rfc1437.ps.gz $z File is in Postscript format. $c Must be decompressed using GUNZIP $s 21K
RECOMMENDATION 6: Staff should include the URL (when readily obtainable) in the 856 $u. When the URL is available, staff may also choose to provide a textual description of location and access information in the LOC $n. If a URL is not readily available, staff should provide textual location and access information in the LOC $n. Because the menu structures of Internet sites are frequently revised, any textual location notes should describe the highest level from which a user could readily find the item. Staff should take care to make notes with location and access information context-neutral, that is, no assumptions should be made about what kind of device is being used or the path through which the user arrived at HOLLIS. The group specifically discussed three situations, here illustrated for the Bryn Mawr classical review:
LOCATION: Networked Resource: Available through HOLLIS Plus. To access: URL:gopher://hplus.harvard.eduTechnical services:
LOC/1:_c: $i net $n Available through HOLLIS Plus. 856/1:7 _: $u gopher://hplus.harvard.edu
LOCATION: Loeb Design: Networked Resource This resource may be accessed using a Graduate School of Design LAN terminal.Technical services:
LOC/1:_c: $i des $c Networked Resource $n This resource may be accessed using a Graduate School of Design LAN terminal. $k netdes
LOCATION: Networked Resource: Available through the Internet. To access: URL:ftp://path.net/pub/journals/bmcr To access: URL:gopher://gopher.lib.virginia.edu/11/alpha/bmcr To access: URL:http://nosferatu.cas.usf.edu/JAC/index.htmlTechnical services:
LOC/1:_c: $i net $n Available through the Internet. 856/1:1_: $u ftp://path.net/pub/journals/bmcr 856/2:7_: $u gopher://gopher.lib.virginia.edu/11/alpha/bmcr $2 gopher 856/3:7_: $u http://nosferatu.cas.usf.edu/JAC/index.html $2 http
One of the fundamental characteristics of networked resources is that they move or disappear frequently, and there is generally nothing to notify a library when this happens. It is very important that the catalog records for these resources be kept accurate and up-to-date, particularly the location and access information. Therefore, libraries must take an active role in verifying the information on a regular basis.
RECOMMENDATION 7: Develop the capability to verify links automatically to the extent possible. In addition, a report should be generated, sorted by library, which each library should use to verify and/or correct the location/access information. This report should be produced quarterly until an automatic verification capability is in place. After a verification mechanism is in place, the report should be produced annually and reviewed for more precise verification than the automatic process can provide. The frequency of these reports can be re-evaluated as the libraries gain experience with cataloging and maintaining cataloging for these resources.
In order to facilitate this maintenance, records for networked resources must be identifiable as a group. Therefore, all cataloging records for networked resources should have a $k in the LOC with the code "netxxx", where xxx is the cataloging library's LOC code. This means that even those resources cataloged under the "$i net" must be traceable to the library which cataloged the resource and is responsible for maintenance of the cataloging. For example, a record for the World Law Index available through HOLLIS Plus and cataloged by the Law Library, would look like this:
LOC/1:_c: $i net $k netlaw
A record for the MLA Bibliography available on the HCL LAN would look like this:
LOC/1:_c: $i wid $c Networked Resource $k netwid
In the period before HOLLIS II, the ability to identify all cataloging records for networked resources could also be used to segregate and copy these records onto a different server where linking would be supported. Being able to identify these records or locations also provides for their removal if a different approach to cataloging networked resources comes out of any of the many projects currently underway at OCLC and elsewhere.
The group agreed that obsolete electronic location and access information should be changed, and that previously valid location and access information should not be kept.
Based on the group's experience, librarians choosing to catalog networked resources will encounter countless situations we have not provided for. We hope that these recommendations will establish a framework within which libraries can gain experience with the bibliographic control of networked resources. We also recognize that education, guidelines, technical support, and mutual assistance will all be necessary for this effort to succeed. Fortunately, there are many forums, such as the Catalogers' Discussion Group, where these issues can be raised and where expertise can be shared.
RECOMMENDATION 8: The Standing Subcommittee on Bibliographic Standards and Policy should extract cataloging guidelines from this paper and coordinate an informational meeting for staff about cataloging networked resources in HOLLIS.
Michael Fitzgerald Ruth Haas Mary Harrington Deborah Kelley-Milburn Joachim Martin Janet Rutan Robin Wendler Hugh Wilburn