Authority Control in the 21st Century: An Invitational Conference
Authority Control in the 21st Century: An Invitational Conference
I will speak today in very practical terms about access control records, and of their structure and uses in future online systems. The prospects for access control are strong, if our conceptual paradigm moves in the direction of the bibliographic control of intellectual works instead of physical (or intangible electronic) items. There are, of course, challenges to be met as we move towards this improved and reorganized bibliographic world, which I will outline in this paper.
The idea of access control as opposed to authority control has been under discussion for nearly twenty years, and there is a small core body of published literature. Michael Gorman outlined his vision (although did not use this specific term) in 1977 at the ALA Information Science and Automation Division conference on the future of the catalog. The first published citation using this term was over a decade later in the proceedings of the 1988 LITA conference in Boston, in an article by Barbara Tillett called "Access Control: a Model for Descriptive, Holding and Control Records." Since 1988, there have been several articles and essays expanding on the idea by both Gorman and Tillett. Access control has also found some recent support abroad by Michael Heaney of the Bodleian Library, who contributed some useful insights in his ITAL article entitled "Object-Oriented Cataloging." And in Tillett's paper at this conference, we have heard the newest developments of this idea.
In this paper, however, I would like to provide some practical insights into the access control model. Using real-life and hypothetical examples, I hope to stimulate further discussion of the benefits and pitfalls for moving in the direction of organizing our online catalogs with access control records (which I find closely tied to the conceptual model of bibliographic relationships and to relational database structures). I believe that it is possible to get to the organizational structure of a work-based catalog from our current record structures. I hope to convince you that doing so is both desirable and feasible. We must consider incremental change to our existing MARC-based system; we do not have the resources to do otherwise.
Why should we do this at all? There are many problems with our current generation of online catalogs that are well known to this audience. Multiple versions of works are ineffectively displayed to users and are not as simple to catalog as they could be; the problem has increased exponentially with electronic publishing. Bibliographic relationships between works are not clear, nor are they consistently displayed. Much redundant information is keyed into both bibliographic and authority records, which is both a waste of resources and ineffective use of computer technology. Worse, work-related data is not consistently keyed into bibliographic records, resulting in incomplete and imprecise user retrievals. A paradigm shift to a work-based catalog would at its best resolve and clarify these current problems.
The access control record is the next generation of the authority record. It can be viewed as a "super authority record" because of the potential it contains for enriched information for indexing. Access control records would be linked both to bibliographic records, to collocate all manifestations of a work, and to other related access control records, to collocate related works. Tillett sees two kinds of control records--names (incorporating both names and titles) and subjects. While this paper is directed towards name and name-title records, many of the concepts could also be applied to subjects.
One of the key concepts with the access control record is removing both the label and the notion of "authority." The access control record is evolved from the current authority record by the fact that it links the variant forms of a name without declaring any one as the "authorized" form. A central concept is that a library or user should be allowed to choose their preferred form of name, or to have displayed a default heading. In Figure 1 some of the essential notions captured in the concept of the access control record are listed.
Figure 1:
Definition of an Access Control Record:
The concept is brazenly radical, and entirely contradicts the whole second part of AACR2, which of course is devoted to painstaking rules for how to construct authorized forms of names and titles. We must recognize that it would require substantive international level discussion and a significant change in one of our fundamental principles to move forward with this concept.
Understanding, then, that in an access control record, all variant forms of a heading are linked together, and any one of those variant forms could be designated by a library or a user as the heading that is displayed, let's look at an example.
Figure 2:
Current Practice for Coding a Name Authority Record:
06 010 n 50018452 07 040 DLC �c DLC 08 100 00 Guillaume, �c de Machaut, �d ca. 1300-1377 09 400 10 Machaut, Guillaume de, �d ca. 1300-1377 10 400 10 Machault, Guillaume de, �d ca. 1300-1377 11 400 10 De Machaut, Guillaume, �d ca. 1300-1377 12 400 10 De Machault, Guillaume, �d ca. 1300-1377 13 400 00 Guillaume de Machaut, �d d. 1377 �w nnaa
This is a familiar format--the current practice for coding a name authority record. This record is for one of the first composers in history whom we know by name--Guillaume de Machaut--a composer difficult for music researchers to find in our online catalogs, despite the cross references.
How might an access control record look different? Here is a hypothetical example.
Figure 3:
Hypothetical Access Control Record:
06 010 n 50018452 07 040 DLC �c DLC 08 400 00 Guillaume, �c de Machaut 09 100 00 Machaut, Guillaume de 10 400 10 Machault, Guillaume de 11 400 10 De Machaut, Guillaume 12 400 10 De Machault, Guillaume 13 400 00 Guillaume de Machaut, �d d. 1377 �w nnaa 14 999 ca. 1300-1377
For dramatic effect, I have switched the 100 field with a 400 field, since many users in my library tend to search under "Machaut." The 1xx field would be considered the "default" display (as opposed to the "authorized form"). The 1xx field might not then be the same in every catalog. This is a key point. I would expect that every national-level access control record would continue to need a 100 field as a "national default" display, which also might be a point of agreement for union catalogs and for users who do not wish to select a default.
I have removed the death dates from each field in this example simply as a reminder that we could consider coding them in a separate field, and to exaggerate the potential difference between current practice and the hypothetical access control record. Why should dates be keyed redundantly with every variant form? Users or libraries would then have the option to display or not display them, as they would other identifying attributes.
As Tillett has suggested, there are many other data elements which could be encoded which would help us uniquely identify an individual. We have learned of some precedent for this in OCLC's development of an in-house corrections file, in which a non-MARC file of non-LCNAF identifier records was created. OCLC has transferred data into this correction file from bibliographic record elements which include: when the author began publishing, the language in which the author published, countries in which the author published, the subject of the publication, the relationship of the author to the publication, and the title of the publication. These or other data elements could be incorporated into access control records. Some other examples include: the individual's occupation; his or her area of research or writing activity; his or her nationality or geographic sphere of influence; the birthplace and deathplace; and dates of activity. Individual data elements (or a combination of data elements) could be selected and displayed by a library to meet their users' needs.
Another enhancement that we could implement for access control records would be additional subfield coding for each variant which would differentiate the various languages and romanizations. A library would then be able to programmatically identify a category of variant forms to suppress or delete, or perhaps even as the default display.
Much of the basic information is already there in our current name authority records. The access control record based on these records need not look too much different, and I think it is possible to get there largely through machine manipulation (not including the enriched data elements, of course). Programs could be written to move dates to a separate new field (if in fact we decide we want to do that), and MARBI could be approached for validation of new fields to hold other enriched data elements for potential display. Programmatic links could be made between bibliographic and access control records.
Bibliographic records would not contain hard-coded headings but would be linked to the access control records. To make this a reality, it is the local online system design and indexing that would require the most substantive development work. What would be different would be the internal model of system design, requiring what we call now "true authority control," which has seldom been implemented. There are of course some commercial systems available now which operate this way, but a far greater number do not.
Figure 4:
Linking an access control record to a bibliographic record:
12 020 0193373955
13 040 DLC �c BTS �d OCL
14 041 0 lat
15 048 zn
16 050 M2010.G95 �b N7
17 100 0 n 50018452
18 240 10 Messe de Nostre Dame
19 245 13 La messe de Nostre Dame. �c Edited by Denis
Stevens.
20 260 London, �b Oxford University Press �c [1973]
21 300 score (32 p.) �c 26 cm.
22 500 "The instrumental participation is limited to
violins, violas, cellos, and trombones, but it
would ... be possible to make use of viols instead
of modern stringed instruments."
23 500 Duration: 27 min.
24 650 0 Masses �x Scores.
In this example of a particular mass of Machaut's, you can see that the text of the 100 field has been replaced by a numerical link to the access control record. For public display, the preferred heading from the access control record would be shown. For technical services display, options include displaying the numerical link, the public (name) display, or perhaps a windowed side-by-side display of bibliographic and access control records.
It is incumbent upon us to lobby and pressure local system designers to move in this direction. Again, I suspect that many of the changes in records can be done through machine manipulation. Could "search and replace" strategies re-code our bibliographic records so that record number links are made, allowing this tremendous flexibility of display?
The notion of not having an authorized form of heading raises many questions. For many names, all variants are roughly equal, and the basis for determining an authorized form is largely artificial. Are there cases, though, where we do want to point the user to a specific form and not support wrong-headed search strategies? Is it appropriate for a user to find the music of Andrew Lloyd Webber when she searches the online catalog under Webber? Is it our responsibility to tell the user that that simply is not his surname? Where do we draw the line? This is one of our challenges.
A bibliographic file linked by pointers to an authority or access control file would be a major change in system design for many vendors, and there should be no illusions about the difficulty or economic impact of such change. However, since the basic record structure and data content is not significantly different, I believe that we can get there from here. What changes is the way the data is manipulated and displayed, not the basic record content.
I now want to move on to the implications of the access control record for name/title control, and specifically for those names and titles representing musical works. The concept of pointing and linking, rather than re-keying headings data manually is extended a step further when talking about name/title relationships. Here access control records will look substantially different. In order to retain the flexibility of name heading selection, links must be made to the name record instead of keyed redundantly into the title record. Let me give an example.
Figure 5:
Current Name/Title Authority Record:
06 010 n 80056438
07 040 DLC �c DLC
08 100 10 Tchaikovsky, Peter Ilich, �d 1840-1893. �t
Shchelkunchik
09 400 10 Tchaikovsky, Peter Ilich, �d 1840-1893. �t Casse-
noisette
10 400 10 Tchaikovsky, Peter Ilich, �d 1840-1893. �t Di�t�ro
11 400 10 Tchaikovsky, Peter Ilich, �d 1840-1893. �t
N�tkn�ppar
12 400 10 Tchaikovsky, Peter Ilich, �d 1840-1893. �t
Nussknacker
13 400 10 Tchaikovsky, Peter Ilich, �d 1840-1893. �t
Schiaccianoci
14 400 10 Tchaikovsky, Peter Ilich, �d 1840-1893. �t
Spargator de nuci
15 400 10 Tchaikovsky, Peter Ilich, �d 1840-1893. �t
Nutcracker
Here we have the current version of the name-title authority record for Tchaikovsky's ballet The Nutcracker (without fixed fields and 670s). It is established with the composer's original Russian title, according to AACR2, with references from forms of the title in various languages.
Figure 6:
Hypothetical Access Control Record (Version A):
06 010 n 80056438 07 040 DLC �c DLC 08 400 10 n 79072979 �t Shchelkunchik 09 400 10 n 79072979 �t Casse-noisette 10 400 10 n 79072979 �t Di�t�ro 11 400 10 n 79072979 �t N�tkn�ppar 12 400 10 n 79072979 �t Nussknacker 13 400 10 n 79072979 �t Schiaccianoci 14 400 10 n 79072979 �t Spargator de nuci 15 100 10 n 79072979 �t Nutcracker
which links to:
Access Control Record for His Personal Name:
06 010 n 79072979 07 040 DLC �c DLC �d DLC 08 100 10 Tchaikovsky, Peter Ilich 09 400 10 Ciaikovsky, Piotr Ilic 10 400 10 Tschaikowsky, Peter Iljitch 11 400 10 Tchaikowsky, Peter Iljitch 12 400 10 Ciaikovsky, Pjotr Iljc 13 400 10 Cajkovskij, P. I. 14 400 10 Tsjaikovsky, Peter Iljitsj 15 400 10 Czajkowski, Piotr 16 400 10 Chaikovsky, P. I. 99 999 1840-1893 ... plus 25 more variant forms!
Here is one option for what an access control record might look like. Again, for dramatic effect, I have changed the 100 field to the English version of the title. Users in my library would much prefer to see this heading rather than a cross reference! This option is a simple substitution of the number of the access control record for the name (as seen on a previous slide). I think this would be entirely possible to do by machine conversion (meaning special programs being written), but as you can see the information is very redundant.
Figure 7:
Hypothetical Access Control Record (Version B):
06 010 n 80056438 07 040 DLC �c DLC 08 430 00 Shchelkunchik 09 430 00 Casse-noisette 10 430 00 Di�t�ro 11 430 00 N�tkn�ppar 12 430 00 Nussknacker 13 430 00 Schiaccianoci 14 430 00 Spargator de nuci 15 130 00 Nutcracker 16 667 n 79072979 �2 LCNAF �4 cmp
Michael Heaney from the Bodleian Library has provided some guidance for another possible option. Here, you will notice that the link to the composer's name is provided only once--for convenience's sake coded in a 667 field. Heaney has also suggested additional coding to show the source of the heading (the subfield 2, LCNAF) and relator codes (subfield 4, "cmp" for composer), utilizing current coding practices in bibliographic records. You can imagine by extension the possibilities for including other names related to the work--the librettists for an opera, the literary work upon which the opera is based, etc. Note also that the basic coding in the record has been changed to 130/430 instead of the 100/400 in the previous example.
Again, I will follow with a mocked-up version of what the bibliographic record might look like.
Figure 8:
Linking an access control record for a work to a bibliographic record:
1 010 52-26400
2 040 DLC �c KSU �d SER �d BGU �d IOM �d OCL
3 028 22 B. B. 47 �b Broude
4 050 0 M1520.C43 �b N8
5 092 784.21556 �b T219s
6 130 0 n 80056438
7 245 14 The nutcracker ballet : �b Casse-noisette ballet :
�erie en 2 actes, op. 71 / �c musique de Peter
Ilich Tchaikovsky
8 246 31 Casse-noisette ballet
9 260 New York : �b Broude, �c c1951
10 300 1 score (511 p.) ; �c 28 cm.
11 500 For orchestra.
12 500 Apparently reproduced from the edition published
by P. Jurgenson, Moscow, 1982?
13 500 Includes English notes on the Russian text.
14 650 0 Ballets �x Scores.
For convenience, the link to the access control record for the work is made through the 130 field. This in turn, as you recall, links to the user-chosen version of the personal name.
It seems to me that this is an area, certainly for music and possibly for literature as well, where libraries have wanted for some time to be able to select and display a different heading than the now-authorized form. Even with our nice reference structure, few users even in academic or research libraries appreciate searching for Tchaikovsky's Nutcracker under the original Russian title Shchelkunchik. It would be wonderful to anticipate the possibility of a library or user-selected heading for music and literary uniform titles using the access control record.
I want to take the concept of the access control record for musical works one step further. I introduce the notion of musical "work" quite deliberately just now, because I would argue that the name/title authority records we have been building in our national and local files for many years are in fact records for works and not for physical items. I am using the term "work" in the Lubetzkian sense of the abstract intellectual concept, with the item as the physical means of conveying the intellectual work.
The extensive LC Music Cataloging Decision for rule 26.4B1 delineates what is done for musical materials as opposed to what is done for books and strengthens the argument for the records for musical "works."
"Generally, the heading referred to should include only the basic uniform title of the work, without additions such as 'arr.,' 'Vocal score,' 'Libretto,' language, etc., even if such additions are used in the uniform title in the bibliographic record for the item being cataloged. If, however, the title being referred from is specific to the arrangement, format, language, etc. brought out by an addition to the uniform title, and the title would not logically be used for a different manifestation of the work, refer to the uniform title with the addition.Bartók, Béla
Duke Bluebeard's castle
search under
Bartók, Béla
Kékszakállu herceg vára"
This is what allows music catalogers to create records such as that for the Nutcracker shown previously, instead of applying the principles of 26.4B1 as done for books.
"26.4B1. Different titles or variants of the title.
Refer to the uniform title from the different titles and variants of the title under which the work has been published or cited in reference sources.....In the case of translated titles, refer to the uniform titles and the appropriate language subheading, when appropriate.
Flaubert, Gustave
Sentimental education
see Flaubert, Gustave
Education sentimentale. English"
This shows the AACR2R rule that differentiates book treatment from the way musical works are treated. I think the examples given (Flaubert for the literary work, and Bluebeard's Castle for the musical work) help to make this distinction especially clear. To refresh your memories, here is how a name-title for a literary work is currently established.
Figure 9:
Current Practice for Coding a Name-Title Authority Record for a Literary Work:
06 010 n 79107982
07 040 DLC �c DLC
08 100 10 Dostoyevsky, Fyodor, �d 1821-1881. �t Prestuplenie
nakazanie. �l English
09 400 10 Dostoevskii, Fedor Mikhailovich, �d 1821-1881. �t
Prestuplenie I nakazanie. �l English �w nnaa
10 400 10 Dostoyevsky, Fyodor, �d 1821-1881. �t Crime and
punishment
The point is that the variant title in English is coded on a different authority record than it would have been for a musical work.
One step that we could take if we want to move in the direction of work-based catalogs would be to synchronize these two approaches, and begin coding all name/title authority references on the same type of headings records. We would first need to achieve clarity on the definition of a work, and, for this example, whether translations are in fact a new work requiring a separate access control record. Based on that decision, these two currently divergent practices could be reconciled.
If we accept that current name/title authority records for music in fact represent abstract works, what might this mean in the access control record environment? I think we have the means to make a major paradigm shift. Here we have a testbed and a mechanism for changing the very foundation of bibliographic control. There has been a surge of interest recently in the literature toward moving to a model of cataloging the work, explicating the various manifestations that represent that work, and defining the bibliographic relationships that connect a work to other related works. Much remains to be done in this area, but there are active researchers and groups who have contributed substantially to inform our basic premises. I look forward to continued developments in this area.
I think we can also make use of the access control record for a work to solve some of the long-standing problems we have had with access to musical materials. One of the directions I would like to see pursued is that of the access control record being considered the "core repository" of information related to a work.
Current techniques of coding work-specific data into the bibliographic records for various manifestations are very inefficient and redundant, and lead to search results with poor precision and recall.
I advocate putting as much data about the work as possible into the access control record, and having it indexed for retrieval. For example, I would like to see instrumentation, date of composition, language, geographic elements, form/genre, even classification and subject headings, coded in and indexed from the access control record. These are all elements that catalogers are now redundantly, sporadically, and inefficiently coding in bibliographic records. In addition, for works with distinctive titles, there would be a place to code opus and thematic index numbers, which is something reference librarians have been asking for for years. In the even further future, I can foresee encoded incipits (the first few bars in music notation) and linked sound files.
Figure 10:
Hypothetical Access Control Record (Enriched):
06 010 n 80056438 07 040 DLC �c DLC 08 045 1 �d 1891 �d 1892 09 047 bt 10 048 oa �a cb02 11 048 oa �a cd02 12 050 M1520 13 092 784.21556 14 430 00 Shchelkunchik 15 430 00 Casse-noisette 16 430 00 Di�t�ro 17 430 00 N�tkn�ppar 18 430 00 Nussknacker 19 430 00 Schiaccianoci 20 430 00 Spargator de nuci 21 130 00 Nutcracker 22 667 n 79072979 �2 LCNAF �4 cmp 23 999 op. 71
In this mocked-up example, I have used fields that are valid in the bibliographic format. Some of you will be familiar with the 045 field (showing a range of dates of composition), the 047 field (showing via coded data that this is a ballet) and the 048 field (showing that the instrumentation of The Nutcracker is for orchestra and two-part women's chorus, and alternatively for children's chorus). I have used these coded fields for convenience, and to emphasize the idea that these attributes are work-related rather than manifestation-related. Should we decide to move in this enriched-coding direction, I think the profession must have a conversation about whether using coded terms, controlled vocabulary terms, or natural language terms is the most efficient and effective. By the way, the 045 field is already authorized for use in authority records.
Let me reiterate, lest it seem like this is a tremendous amount of work, that this is an efficient thing to do. It makes sense to code and index these elements in an access control record, where they will be retrieved consistently for all manifestations. This is an intellectual process appropriate for catalogers to do once. Access control records could be enriched in this way through projects and special initiatives. The potential for vastly improved music retrieval alone with this kind of a data-rich file are enormous; many powerful kinds of intersected searches would be possible for scholars and other users.
Access control records, then, could become the intellectual locus of the online catalog and the basis for a work-based online catalog. They would allow for enriched information about individuals and works which would result in flexible but consistent indexing and retrieval. Access control records allow for less rigidity (or authority) in displays, and the user or library could select the default display form.
I foresee that this is where catalogers will be spending their time, in coding appropriate data for the work including variant forms in the access control record; in forging links between related access control records and between access control and bibliographic records, and in determining and coding the relationships of various manifestations. Bibliographic records will become smaller and simpler, and describing an item in hand, a physical or digital manifestation, could have a much smaller role in the cataloging process.
I find this an extremely attractive model for musical, art and literary works. The implications are not as clear for the sciences, social sciences, etc. The workload for establishing access control records for all works--once we clearly understand what we means by a work--is enormous. The transition to this type of model will require an investment both in terms of vendor labor and library cataloging staff labor.
In sum, there has been a groundswell in the profession toward rethinking the conceptual foundations of bibliographic control. It appears that a model which distinguishes works and items, and links and explains various relationships might be a viable possibility. We have an opportunity to utilize authority records for musical works as a testbed for just such a model. It is possible to get there from here with incremental change, but there are many questions to be answered. We need to continue to think and talk and move forward.
1. Gorman, Michael. "Cataloging and the new technologies." In: The Nature and Future of the Catalog: Proceedings of the ALA's Information Science and Automation Division's 1975 and 1977 Institutes on the Catalog. Edited by Maurice J. Freedman and S. Michael Malinconico. Mansell: Oryx, 1979. p. 127-36.
2. Tillett, Barbara. "Access control: a model for descriptive, holding and control records." In: Convergence: proceedings of the second national Library and Information Technology Association, October 2-6, 1988, Boston, Massachusetts. Chicago: American Library Association, 1990, p. 48-56.
3. Tillett, Barbara. "Bibliographic Structures: The Evolution of Catalog Entries, References, and Tracings." In: Conceptual Foundations of Descriptive Cataloging, edited by Elaine Svenonius. San Diego: Academic Press, 1989, p. 149-165 ; Gorman, Michael. �After AACR2: the Future of the Anglo-American Cataloging Rules.� In: Origins, Content, and Future of AACR2 Revised, edited by Richard P. Smiraglia. Chicago: American Library Association, 1992, p. 89-94; Tillett, Barbara. �Future Cataloging Rules and Catalog Records.� In: Origins, Content, and Future of AACR2 Revised, edited by Richard P. Smiraglia. Chicago: American Library Association, 1992, p. 110-118; Tillett, Barbara. �21st Century Authority Control: What is it and How do we get there?� In: The Future is Now: Reconciling Change and Continuity in Authority Control; Proceedings of the OCLC Symposium, ALA Annual Conference, June 23, 1995. Dublin, Ohio: OCLC Online Computer Library Center, 1995, p. 17-21.
4. Heaney, Michael. "Object-Oriented Cataloging." Information Technology and Libraries, Vol. 14, no. 3 (September 1995), p. 135-153.
5. IFLA Study Group on the Functional Requirements of Bibliographic Records. Functional Requirements of Bibliographic Records. Unpublished draft, July 31, 1995.
6. Smiraglia, Richard P. �Authority Control and the Extent of Derivative Bibliographic Relationships.� Ph.D. diss., Graduate Library School, University of Chicago, 1992; Tillett, Barbara B. �Bibliographic Relationships: Toward a Conceptual Structure of Bibliographic Information used in Cataloging.� Ph.D. diss., Graduate School of Library & Information Science, University of California, Los Angeles, 1987; Vellucci, Sherry Lynn. �Bibliographic Relationships Among Musical Bibliographic Entities: A Conceptual Analysis of Music Represented in a Library Catalog with a Taxonomy of the Relationships Discovered.� D.L.S. diss., School of Library Service, Columbia University, 1995; Yee, Martha M. �Manifestations and Near-Equivalents: Theory, with Special Attention to Moving-Image Materials.� Library Resources & Technical Services, Vol. 38, no. 3 (July 1994), p. 227-255.
Return to Proceedings Home Page
Advanced Search | Careers at OCLC | Feedback | Privacy Policy | ISO 9001 Certificate | ©2003 OCLC Online Computer Library Center, Inc. |