challenges of using metadata in a library setting: the collection and management of electronic links...

11
Challenges of using metadata in a library setting: the Collection And Management of Electronic Links (CAMEL) project at Oregon State University Kyle Banerjee* Oregon State University, 3769 Valley Library, Corvallis, OR 97331-4501, USA Abstract In late 1998, the Valley Library at Oregon State University embarked on a project to provide access to electronic resources in its catalog. This project, known as CAMEL (Collection And Management of Electronic Links), was started because students and faculty were increasingly relying on Web-based resources, but inadequate access was being provided to these resources. A small team was charged with developing a process that would help achieve the goals of better identifying and providing access to these resources in the online catalog. Because the quantity of information being provided in electronic form is expanding very rapidly, this team also investigated the possibilities of using metadata as well as other tools to reduce the need to manually create records for electronic resources. This paper outlines some of the social, technical, and organizational challenges encountered during the CAMEL project. Further information about CAMEL may be found at http://osu.orst.edu/dept/library/camel/. © 2000 Elsevier Science Ltd. All rights reserved. Keywords: Cataloging electronic resources; Metadata; Academic libraries 1. Introduction Metadata appeals to many people on an intuitive level and it can potentially alleviate proven needs in information retrieval. However, that does not mean that metadata will necessarily play an important role in the future. There are many logical tools that have been * Corresponding author. E-mail address: [email protected] (K. Banerjee). Library Collections, Acquisitions, & Technical Services 24 (2000) 217–227 1464-9055/00/$ – see front matter © 2000 Elsevier Science Ltd. All rights reserved. PII: S1464-9055(00)00130-5

Upload: kyle-banerjee

Post on 05-Jul-2016

212 views

Category:

Documents


0 download

TRANSCRIPT

Challenges of using metadata in a library setting: theCollection And Management of Electronic Links

(CAMEL) project at Oregon State University

Kyle Banerjee*

Oregon State University, 3769 Valley Library, Corvallis, OR 97331-4501, USA

Abstract

In late 1998, the Valley Library at Oregon State University embarked on a project to provideaccess to electronic resources in its catalog. This project, known as CAMEL (Collection AndManagement of Electronic Links), was started because students and faculty were increasinglyrelying on Web-based resources, but inadequate access was being provided to these resources. Asmall team was charged with developing a process that would help achieve the goals of betteridentifying and providing access to these resources in the online catalog. Because the quantity ofinformation being provided in electronic form is expanding very rapidly, this team alsoinvestigated the possibilities of using metadata as well as other tools to reduce the need tomanually create records for electronic resources. This paper outlines some of the social, technical,and organizational challenges encountered during the CAMEL project. Further information aboutCAMEL may be found athttp://osu.orst.edu/dept/library/camel/.© 2000 Elsevier Science Ltd. Allrights reserved.

Keywords:Cataloging electronic resources; Metadata; Academic libraries

1. Introduction

Metadata appeals to many people on an intuitive level and it can potentially alleviateproven needs in information retrieval. However, that does not mean that metadata willnecessarily play an important role in the future. There are many logical tools that have been

* Corresponding author.E-mail address:[email protected] (K. Banerjee).

Library Collections, Acquisitions,& Technical Services 24 (2000) 217–227

1464-9055/00/$ – see front matter © 2000 Elsevier Science Ltd. All rights reserved.PII: S1464-9055(00)00130-5

around for a very long time such as Esperanto and the Dvorak keyboard. Neither of these hasgained a significant level of acceptance despite a need for better communication and a wayto type faster without increasing the risk of carpal tunnel syndrome. The bottom line is thatimplementing new ways of doing things is difficult.

Significant human, organizational and technical issues must be addressed before a librarycan implement a mechanism that relies on metadata. Information providers need to beinformed about metadata and convinced to create it. New tools that employ metadata mustbe integrated with existing library access mechanisms to minimize confusion to library staffand users. Standards need to be implemented to ensure consistency. Systems need to be ableto properly interpret it. Also, procedures must be developed to address maintenance problemsbrought about when relatively unstable electronic resources are deleted, moved, renamed, orchanged.

In the fall 1998, Oregon State University initiated a project that sought to explore manyof these issues. This project, known as CAMEL (Collection And Management of ElectronicLinks), attempted to use metadata to facilitate access to networked university resourcesthrough the online catalog. The sections that follow outline some of the social, technical, andorganizational challenges encountered during the CAMEL pilot project. Further informationabout CAMEL may be found athttp://osu.orst.edu/dept/library/camel/.

2. The CAMEL project

The CAMEL project was started because students and faculty were increasingly relyingon Web-based resources, but inadequate access was being provided to these resources.Despite the fact that the library has always relied heavily on faculty recommendations foracquiring print resources, no formal mechanism for recommending useful Web-based re-sources existed. A small team was charged with developing a process that would helpachieve the goals of better identifying and providing access to these resources in OASIS, theonline catalog at Oregon State University. Because the quantity of information beingprovided in electronic form is expanding very rapidly, this team also explored using metadataand other tools to reduce the need to manually create records for electronic resources.

2.1. How is CAMEL different from other Internet cataloging projects?

A number of initiatives such as OCLC’s CORC (Cooperative Online Resource Catalog)project and the Internet Scout projects are currently devoted to improving access to Internetresources. Although these projects and CAMEL all rely on manually created databaserecords to provide access to Internet resources, they differ considerably in terms of focus. Forexample, CORC’s primary aim is to facilitate the exchange of cooperatively created meta-data among libraries. CORC uses the WorldCat model as a basis for sharing records ofInternet resources, and it explores the development of automated tools to aid in recordcreation. CORC also uses emerging standards such as XML as a basis for storing andexchanging information. Internet Scout concentrates on disseminating information aboutInternet resources discovered by its staff. As such, it functions much like a readers advisory

218 K. Banerjee / Libr. Coll. Acq. & Tech. Serv. 24 (2000) 217–227

service, though the project maintains a permanent database of resources likely to be ofinterest to the research and educational communities.

CAMEL mainly seeks to identify valuable Internet resources by improving communica-tion between librarians, faculty, and subject specialists. CAMEL also seeks to explore thepossibility of shifting some of the responsibility for describing and maintaining records tousers. In short, CAMEL aims to reduce dependence on library professionals to identify anddescribe appropriate resources. Generally speaking, the CAMEL project differentiates itselffrom other efforts to include Internet resources in the online catalog in four importantrespects:

1. University faculty identify resources that are then screened by subject specialists.

One of the main goals of the CAMEL project is to increase communication betweenfaculty and subject specialists so that the process for identifying Internet resources approx-imates the process used to identify books, journals, films, etc. in traditional formats. Mostprojects aimed at including Internet resources in the online catalog rely on librarians orstudents with subject knowledge to find resources. Faculty participation is a critical aspect ofOregon State’s CAMEL project.

2. Faculty who submit resources help identify appropriate access points that willfacilitate discovery.

To submit a resource, faculty members must supply a title, description, and a URL. Theyare also strongly encouraged to submit authors’ names, keywords, and other relevantinformation. Originally, it was hoped that this information could be used to automaticallygenerate metadata that would facilitate searching. In practice, however, this information hasbeen mostly used to help librarians determine appropriate access points. The reasons forabandoning the goal of generating metadata automatically will be explained later.

3. Faculty who identify resources share some of the burden for maintaining access tothese resources.

Any person who submits a resource to CAMEL is asked to supply an e-mail address. Theexpectation is that faculty are likely to submit resources in which they have some directinterest. If questions later arise about the resource (i.e., it disappears or changes), the personwho submitted the resource may be asked if they can supply information that will help thelibrary.

4. The CAMEL project uses a separate Internet-based database that serves as a stagingarea for verification of resources to be included in the online catalog.

There are two components to CAMEL: 1) a Web-based database; and 2) records inOASIS, the online catalog. When a faculty member submits a resource, a subject specialistdetermines whether the library should provide access to the resource. If so, the informationsupplied by the submitter is edited if necessary and the resource is immediately added to theWeb-based database. At this time, a decision is made as to whether the resource should beincluded in OASIS. Catalogers are notified via email of approved resources and a MARCrecord is created in accordance with normal cataloging rules. OASIS control numbers are

219K. Banerjee / Libr. Coll. Acq. & Tech. Serv. 24 (2000) 217–227

stored in the CAMEL database to simplify synchronization of records in the Web-baseddatabase and online catalog.

CAMEL uses the online catalog in conjunction with a Web-based database for tworeasons. First, library users expect to find library materials in the online catalog.Secondly, most library catalogs lack many features that are useful for creating andmaintaining a database of online resources. For these reasons, CAMEL provides accessto online resources via the catalog, but uses a separate (but linked) database to facilitatemaintenance. CORC also maintains a separate database, but it is not directly related toOCLC’s main bibliographic database. As such, OCLC must use parallel maintenancemechanisms and inconsistencies between records in the two databases can result.

5. The collection development policy focuses on resources and publications created atOregon State University.

One of the first tasks of the CAMEL team was to identify what type of Internetresources should be included in the CAMEL database and OASIS catalog. For a numberof reasons, the goal of providing and maintaining access to all types of resources on theInternet was dismissed as undesirable as well as unrealistic. Much of the information onthe web is dynamic in nature and has only transitory value [1]. Moreover, the nature andquantity of information on the Internet is such that experienced professionals usingcontrolled vocabularies can only hope to organize small and well-defined collections ofresources. Because the Valley Library is committed to providing access to resourcesproduced at Oregon State University, and these resources constitute a discrete andmanageable set, the decision was made to focus solely on resources created by OSUfaculty and departments.

2.2. How does the CAMEL project use metadata?

One of the initial motivations of the CAMEL project was to identify ways to use authoror subject specialist generated metadata to dramatically reduce manual processing by cata-loging staff. Originally, the CAMEL team wanted to use a Web-based template to convertuser-supplied information into Dublin Core (DC) metadata and perhaps the skeleton of aMARC record. However, once the first template was designed, it was apparent that it wasunreasonable to expect diverse users to fill out complete and correct DC information. Someelements such as publisher seemed ambiguous, whereas others might not make sense at allto people submitting resources.

It is important to note that although CAMEL focuses on OSU-produced resources, facultymembers may submit resources that they did not create themselves (e.g., departmentalpublications). People cannot embed metadata in works that they do not control, so thedecision was made to produce a form designed primarily to notify a library subject specialistof the existence of a particularly useful resource and to supply a few basic access points.Because one of the objectives of CAMEL is to use the existing online catalog to provideaccess to electronic resources, the metadata used by the CAMEL project is very similar tothat used in a catalog record.

In the end, a very simplified form containing author, title, location, description,

220 K. Banerjee / Libr. Coll. Acq. & Tech. Serv. 24 (2000) 217–227

keywords, resource type, and name of the person submitting the resource was imple-mented (see Fig. 1). All other fields necessary for inclusion in the online catalog areadded manually. Currently, the true role metadata plays in the CAMEL project is roughlyanalogous to the role CIP data plays in printed materials—to help the database main-tainers determine appropriate access points more quickly rather than to provide completeinformation.

3. Human and organizational issues

Human and organizational considerations significantly limited the ability of theCAMEL project to employ metadata that could be generated automatically as well as bycasual database users. To implement any system that uses metadata, people as well asorganizational units must be informed and convinced to use new tools and procedures.For this reason, it is critical to have a plan for addressing the social aspects of any newmechanism. The following sections outline some of the challenges encountered by theCAMEL team.

Fig. 1. CAMEL resource submission form from http://osu.orst.edu/dept/library/camel/submit.cfm.

221K. Banerjee / Libr. Coll. Acq. & Tech. Serv. 24 (2000) 217–227

3.1. Convincing faculty to submit resources to the CAMEL database

The first task faced by the CAMEL team was to convince faculty to fill out a Web-basedtemplate designed to alert library subject specialists to the existence of important Internetresources. At the outset of the project, many faculty regarded CAMEL as simply anotherWeb search engine and felt the project was unnecessary. Moreover, it was necessary for theteam to communicate what types of resources CAMEL sought to provide access to—i.e.,genuinely useful resources with some lasting value as opposed to resumes and home pages.

To address these problems, library subject specialists contacted faculty by phone, email,and in person. Generally, faculty response to the project was tepid, but most became moresupportive when they learned that the items they submitted might receive full level catalog-ing and be made available through the online catalog. Anecdotal evidence suggests thatinterest in adding Internet resources to the catalog was strong for two main reasons. First, thecatalog is the main access mechanism for all library resources, so items represented there aremore likely to be found. Secondly, few people were interested in learning a new accessmechanism if the old ones (i.e., the catalog) must still be used.

Relying on faculty to submit resources met with mixed success. On one hand, somefaculty members were excited about the project and submitted a number of resources.Moreover, virtually all resources submitted have been of high quality and some facultycontinued to submit resources weeks after being contacted. However, participation fromfaculty has been uneven overall. Faculty who serve off-campus populations have beenenthusiastic about supporting the project, but others have not contributed a significantnumber of resources.

3.2. Convincing information providers to use standardized metadata

Once faculty were convinced to submit resources using the Web-based template, theCAMEL team then focused on communicating what sort of metadata belonged in variousfields. Metadata must be entered consistently if they are to be useful [2]. To achieveconsistent entry of metadata, people have to agree upon what type of information belongs inthe metadata and what they hope to accomplish by putting it there.

Attaining this agreement is extremely difficult as a practical matter. There are many effortsaimed at organizing Internet resources. User communities with diverse interests back theseefforts to provide access to diverse resources, and each “standard” typically is defined interms that reflect the interests of a particular user community [3]. For this reason, theCAMEL team focused only on metadata elements that were as universally applicable aspossible, using the DC as a model.

Adding consistent metadata requires significant time and effort. Although some metadatamay be generated automatically, only humans can identify many important relationships indata—especially when these relationships extend beyond the item being examined [4]. Toachieve consistency, external standards must be followed. Applying these standards can bedifficult because standard authorities and practices may lack the expressive power to definea resource or they may require an element that is not present in the work at hand [5].

When the CAMEL project started, it was hoped that most submitted records would require

222 K. Banerjee / Libr. Coll. Acq. & Tech. Serv. 24 (2000) 217–227

little manual editing. However, as the project developed, it became clear that significant laborresources were necessary to make the submitted records useful. There was a great deal ofvariability in how users filled out the CAMEL submission form. Content, syntax, capitali-zation, and spelling varied in the free-text fields, though the problems occurred more oftenin some fields than others. For example, the titles and descriptions people supplied typicallyrequired little editing.

On the other hand, the quantity, nature and granularity of user submitted keywords varieddramatically in relation to the resources they described. For example, a submission for aresource containing information on literally hundreds of different kinds of plants includedkeywords for dozens of the specific plants listed. In another case, the phrase “directmarketing” was used in the keywords for a site associated with managing small farms eventhough direct marketing had very little to do with the site. In other cases no keywords orfairly meaningless general terms were supplied. Because of the inconsistency in the keywordfield, some members on the CAMEL team felt that the value of the field was questionable.In practice, CAMEL neither uses keywords as an access point nor as an aid for identifyinggood access points. However, as the database grows and the project evolves, the value of thisfield may become more apparent.

Librarians should be aware that when people supply terminology describing a resource,they often think only in terms of providing access to the resource at hand—not in termsrelating the resource to other items in the database. Even when people think in terms of thedatabase as a collection, it is difficult to find appropriate terminology to describe a resourcebecause there is some ambiguity over what constitutes the collection and what constitutes themetadata describing it in the digital realm [6]. Another problem is that people supplymetadata for different reasons—some people may be more interested in achieving their ownfinancial or emotional objectives than in helping users find what they need [7]. As a resultof these problems, widely varying terminology is found in the CAMEL keyword field.

3.3. Creating metadata that is useful for staff

Metadata is a tool, not a solution to problems. Just as high tech woodworking and machinetools are potentially useless (or even harmful) in the hands of unskilled people, people needto understand what metadata does and develop certain skills to make use of it. There arealready a great variety of information resources and access mechanisms in the library.Introducing new ones undermines the goal of providing integrated access to all the library’sholdings and requires people to learn new skills.

CAMEL resources are generally included in OASIS, so library staff and patrons do notneed to acquire new skills or use new tools to find these electronic works. However,integrating CAMEL resources into the online catalog effectively required CAMEL metadatato conform to existing cataloging practices. Although this decision undoubtedly simplifieddiscovery of electronic resources for many library staff and patrons, it also ensured thatsearching options for full-text Internet resources would be no better than those used forfinding paper resources. Although good access is provided to CAMEL resources that areincluded in OASIS, the quality of access to CAMEL resources that are not added to thecatalog is relatively low. Unstable resources as well as works that do not meet certain

223K. Banerjee / Libr. Coll. Acq. & Tech. Serv. 24 (2000) 217–227

selection criteria may be judged appropriate for the CAMEL database, but not OASIS. Inthese cases, the user must search the CAMEL database separately. In practice, few peoplesearch the separate CAMEL database.

4. Technical issues

Obviously, it is very important to have a plan for addressing the needs of the people whowill be working directly with metadata before embarking on a large-scale project. However,implementing new ways of collecting, storing and retrieving information requires far morethan cooperation on the part of various groups and individuals. Many technical issues mustbe addressed to ensure that resources will be available in the future and that compatibilitywith present information and access mechanisms is maintained. The CAMEL team spent aconsiderable amount of time overcoming technical difficulties.

4.1. Finding local expertise to support the project

As a practical matter, it is difficult to find people with the specialized skills necessary toevaluate, implement, and maintain systems that exploit metadata. When the project wasoriginally proposed, approval had been given for using ROADS (Resource Organization andDiscovery in Subject-based Services) software. This decision was based on a number offactors, the most important of which were that the tool was already in use by a number oflibraries, its databases could be linked, and it included Z39.50 functionality.

However, finding someone who could dedicate time to the project with sufficient UNIXadministration skills proved difficult. Moreover, Oregon State University computing policydiscourages the installation of software that is not locally supported because of administra-tive and security concerns. For these reasons and others, the project was changed so that itwould be based on a Cold Fusion database instead. As is the case in many places, computingsupport resources at OSU were already stretched thin, so local support was not availablebeyond a minimal level. Even though Cold Fusion is locally supported, it took a considerableamount of time to achieve a base level of functionality. Considerable discussion was devotedto determining which fields were needed, how tables containing these fields would bestructured, limitations on the format or size of a field, etc. Then, it took some time to developcode that would make the forms used for adding resources to and maintaining the databaseperform as expected. The CAMEL project would have faced significant technical challengeseven if local expertise were not an issue.

4.2. Compatibility with present access mechanisms and data

One of the most difficult aspects of introducing metadata is doing so in a manner that iscompatible with existing access mechanisms. The catalog is the primary access point to thevast majority of library resources. Moreover, MARC has become the accepted standard forexchange of library-maintained bibliographic information and has heavily influenced storageand display of information. Given that there is no realistic possibility that the vast majority

224 K. Banerjee / Libr. Coll. Acq. & Tech. Serv. 24 (2000) 217–227

of existing library resources will be recataloged, any new system that relies on metadata mustbe able to take advantage of the enormous amount of data already stored in MARC records[8].

However, the publishing model on which MARC is based does not apply readily toInternet resources—it is often awkward to encode information that is required in MARCfixed and variable length fields such as publication date, publisher, or format. What consti-tutes an individual resource is sometimes unclear, many resources are updated frequently orcontinuously, and one information resource may consist of objects in many formats. Internetresources are dynamic by nature and have complex relationships with other resources. Theyalso differ from print resources in how they are used and how people search for them. Forthese reasons and others, it is difficult to create and maintain useful access points for Internetresources in a MARC record despite the existence of special fields to describe remoteelectronic resources.

4.3. Lack of standards

There are no widely accepted metadata standards yet. Some well-known efforts haveattracted a great deal of interest. However, use of these standards is infrequent and incon-sistent because different interest groups favor competing metadata standards, and none ofthese has achieved a critical following of support [9]. The DC has generated particularexcitement in the library community because of its extensibility and the fact that its elementstransfer relatively easily to MARC. Despite its popularity, the DC defines only a minimal setof elements and it has developed very little over the past few years. The DC is a very flexiblestandard, but its goals of simplicity, extensibility, and interoperability among collections andsystems are very difficult to attain because they sometimes work at cross-purposes [10]. Asa practical matter, few documents contain DC metadata at this point, and it remains to beseen whether it will receive enough support to be viable in the long term. Even if suchsupport materializes, it seems unlikely that the application, structure, and use of DC elementswill be consistent across various user communities [11]. In short, there is no guarantee thatmetadata generated today according to DC or any other emerging standard will be useful forproviding access to documents in the future.

4.4. Software tools

Current software tools that might potentially be used to create metadata are in a veryprimitive stage of development. Most of them (including the CAMEL submission andmaintenance forms) are essentially templates that format user input correctly or scripts thatmap already-structured information into appropriate fields. OCLC’s CORC project doesprovide more sophisticated tools that begin to address issues such as authority control andcontent analysis, but even these tools are fairly rudimentary in terms of their functionality.There are a number of reasons for this, but one of them is that computer-based tools havedifficulty determining what is important to people. Some types of metadata such as date ofcreation or the extent of an item can be generated easily. Other types such as subject orrelationship to other documents are best identified by people because they cannot be readily

225K. Banerjee / Libr. Coll. Acq. & Tech. Serv. 24 (2000) 217–227

derived by software from the structure and content alone. Tools that can search metadata arereasonably sophisticated. Several search engine companies develop products that can workwith metadata. Before purchasing a particular tool, it is important to make sure that it can beconfigured to recognize metadata in the documents being searched. There is little point ingenerating metadata that will simply be ignored or incorrectly interpreted by the tools.

4.5. Maintenance

Digital resources are notoriously unstable. Because of this, any system that utilizesmetadata should be able to address problems caused by deletion, moving, renaming, orchanging of resources. Automated library systems typically cannot perform these functionsyet and questions arise as to how access to these resources will be maintained and who willbe responsible for maintaining this access. The CAMEL project has not been in existencelong enough for meaningful evaluation of its maintenance provisions. The CAMEL projectcontains two mechanisms designed to facilitate maintenance. The first of these is that emailaddresses of people submitting resources are retained. Once each year, these people will beautomatically mailed a short message listing resources they submitted that have beenmodified or deleted. The hope is that the original submitters will be willing to verify recordsfor the resources they submitted and fill out forms detailing any necessary changes. Thesecond maintenance provision is that the bibliographic control numbers of resources thathave been added to the OASIS online catalog are stored in the CAMEL record. Thisinformation links the CAMEL and OASIS records together.

5. Conclusions

On the surface, the problem of how to implement metadata in a library setting appears tobe mostly technical in nature. However, social and organizational issues play a critical rolein determining whether a particular metadata initiative is successful. For this reason, one ofthe lessons learned by the CAMEL project is that those who wish to introduce new accessmechanisms should collaborate as broadly as possible with individuals and organizationalunits.

At this point in time, the CAMEL project is in an early stage of development, so it is onlypossible to draw preliminary conclusions. Initial results suggest that the CAMEL pilotproject has enjoyed limited success. The quality of resources submitted has been uniformlyhigh and faculty have helped librarians at Oregon State University identify a number oflocally produced resources that should be in the catalog. It is hoped that CAMEL’s main-tenance provisions will prove significantly more useful and cost effective than link checkers.However, it is still too early to determine whether retaining contact information fromindividuals submitting resources will facilitate maintenance of the database.

CAMEL made only modest use of metadata to reduce manual processing of materials.Generally speaking, the metadata provided by faculty experts were not significantly better forcataloging purposes than automatically generated metadata. For various reasons, even sub-ject experts do not submit consistent metadata in terms of granularity, syntax, quantity, or

226 K. Banerjee / Libr. Coll. Acq. & Tech. Serv. 24 (2000) 217–227

content. Consequently, CAMEL’s goal of having faculty and subject experts provide infor-mation that can be used to quickly identify and describe Internet resources has been onlypartly fulfilled.

Significant developments have occurred over the past few years with regards to tools andmethods for working with metadata. The DC standard is gaining support inside and outsidethe library community. However, neither the DC nor any other standard has achieved acritical level of use in libraries. It is still unclear which metadata tools and procedures willbenefit libraries most in the long run. Even large cooperative efforts have experienceddifficulties generating large amounts of high quality metadata quickly. Six months afterOCLC’s CORC project was launched, 80 participating institutions had collectively entereda total of only about 2000 original records [12]. Libraries still have relatively little experienceproviding access to remote electronic resources. However, with further experimentation itshould be possible to identify solutions that utilize metadata to enhance access to digitalresources for the benefit of library users.

References

[1] Day M, Heery R, Powell A. National bibliographic records in the digital information environment: metadata,links and standards. Journal of Documentation 1999;55:17.

[2] Fayyad U, Uthurusamy R. Data mining and knowledge discovery in databases. Communications of theACM 1996;39(11).

[3] Vellucci SL. Options for organizing electronic resources: the coexistence of metadata. ASIS Bulletin1997;24(1). http://www.asis.org/Bulletin/Oct-97/vellucci.htm.

[4] Marshall CC. Making metadata: a study of metadata creation for a mixed physical-digital collection.Proceedings of the Third ACM Conference on Digital Libraries. In Witten I, Aksyn R, Shipman F, editors.Digital libraries 98: the third ACM Conference on Digital Libraries. New York: Association for ComputingMachinery, 1998. p. 162.

[5] Ibid., p. 168[6] Ibid., p. 166[7] Milstead J, Feldman S. Metadata: cataloging by any other name. Online 1999;23(1):29.[8] Weinstein PC. Ontology-based metadata: transforming the MARC legacy. Proceedings of the Third ACM

Conference on Digital Libraries. In Witten I, Aksyn R, Shipman F, editors. Digital libraries 98: the thirdACM Conference on Digital Libraries. New York: Association for Computing Machinery, 1998. p. 254.

[9] Milstead, p. 40[10] Hakala J, et al. The Nordic Metadata Project: Final Report. http://linnea.helsinki.fi/meta/nmfinal.htm.[11] Vellucci SL. http://www.asis.org/Bulletin/Oct-97/vellucci.htm.[12] Hickey TB. Conversation at OCLC Update Breakfast. New Orleans, LA. June 26. 1999.

227K. Banerjee / Libr. Coll. Acq. & Tech. Serv. 24 (2000) 217–227