Authority Control in the 21st Century: An Invitational Conference
It would be a disgrace in this day and age to abandon the collocating and finding objectives of the catalog on the grounds that they are too expensive to fulfill. Previous speakers have shown how, in this day and age, constructive solutions, not heretofore possible, are being developed to address the cost problems that threaten the integrity of our catalogs. These include automatic error correction of authority headings, cooperative agreements for the creation and maintenance of records, and rethinking conceptual foundations to create more effective and efficient rules for authority control. My paper falls into the conceptual foundations category. It concerns the relative roles of added entries and cross references in fulfilling the objectives of the catalog. The question it raises is what information belongs on bibliographic records and what information belongs on authority or access control records.
In Panizzi's day, when catalogs were in book form, no provision was made for added entries. Added entries could not be an efficient method of providing access in an environment where each entry for a book had to be written out by hand. Thus, a full or main entry for a book was created; the heading on this entry, i.e. the main entry heading, was the name of the primary author of the book, or if the book lacked an author, its title. Other access points to the book took the form of cross references.
When the Library of Congress printed card system was instituted at the beginning of the century, added entries became viable. Gradually they were substituted for cross-references. The 1908 Rules instructed that added entries or references be made for secondary authors. Cutter also said that for secondary access either added entries or cross references would do and observed that with the unit card system using added entries rather than references could provide more information for less work. The cost-savings of substituting added entries for cross-references were obvious; however, the implications for collocation were not. Even today these implications are not widely comprehended. Still today the relative roles of added entries and cross-references are not clearly delineated. For instance, AACR2 contains the following rule:
26.6A If a number of added entries under the same heading are required, optionally replace them by appropriate references.
Hamlet
For editions of this work, see Shakespeare, William.
Hamlet.
{The injunction here is to not make title proper added entries for editions of Hamlet.}
The interchangeability of-- or confusion in -- the relative roles of added entries and cross-references means the information on a bibliographic record is not clearly differentiated from that on an authority record. This constitutes a conceptual failing and a practical problem.
One consequence of not distinguishing the roles of added entries and cross-references is redundant entry. There are cases where AACR2 prescribes the same information be given in the form of an added entry and a cross-reference. For instance, a French movie has the title proper: L'Amour a la mode . For the English version of the film AACR2 contains a rule to provide an added entry for the English translation of the title proper and also for a cross-reference from this title proper to the uniform title:
Bibliographic record:
Love a la mode {added entry for title proper 25.2E}
[L'Amour a la mode] {uniform title}
Authority record:
Love a la mode
See L'Amour a la mode {cross reference 26.4}
Note that the attribute: title proper; appears on the bibliographic record both in its French and its English form. Prima facie it would seem that the information on bibliographic records should be attributes of the items being described and that each attribute should be represented once only. The information on authority or access control records should be relationships between attributes, including the equivalence relationship that links alternative names of attributes.
The argument might be made that the title proper added entry in the above example is needed to fulfill the finding function of the catalog. A user may want only a subset of the work represented by the film, viz. those copies of the film bearing the title proper in English. Note that in an manual environment, eliminating the added entry title proper would deliver to the user the work, i.e. records for all manifestations of a work, rather than a subset of it. This might not be so bad; indeed Lubetzky would argue it might be better.
However, in an online system a separately indexed field is not needed to provide access to the subset of a work. Making a separately indexed field for a title proper added entry is equivalent to constructing an uncontrolled access point. Uncontrolled access points of course have a place in online catalog but, from a cost point of view, such access is better handled through keyword searching than through a separately indexed field.
In the example just given the uniform title retrieves a work set and the title proper a subset of that work. In the following example, where a work appears in two languages, an added entry on a bibliographic record and an a cross reference on an authority record lead to identical sets:
A search under Abbott Offre de transport maritime will retrieve:
Abbott, R. Offre de transport maritime de ligne reguliere au Canada.
$l English
See
Abbott R.
The supply of liner shipping to Canada
as well as:
100 1- Abbott, R.
245 14 The supply of liner shipping to Canada / $c by R. Abbott, Z. Mockus, N. Farinaccio.
246 15 Offre de transport maritime de ligne reguliere au Canada
Another case where a cross reference and an added entry lead to identical retrieval sets is where a person is represented in a catalog in both an official and a private capacity.
Added Entry: Nixon, Richard M. (Richard Milhous, 1913-
The Authority Record with the heading:
110 United States. President (1969-1974 : Nixon)
contains either:
500 Nixon, Richard M. (Richard Milhous), 1913-
or:
663 Here are entered works for Richard Nixon acting in his official capacity. For other works, see Nixon, Richard M. (Richard Milhous), 1913-
In a time when the high costs of catalog construction weigh heavily on library administrators' minds, so much so that a solution being proposed is to reduce access, shouldn't a first consideration be to reduce redundant access? The following example taken from the Rule Interpretation for Title Added Entries illustrates a case of redundancy where the redundant entries are all on the bibliographic record.
100 1- Shakespeare, William, $d 1564-1616
240 10 Midsummer night's dream
245 10 Shakespeare's A midsummer night's dream
246 30 Midsummer night's dream
The Rule Interpretation quite rightly points out that the 240 and 246 fields perform different functions. But again the questionis whether there is a need in an online catalog -- and even in a card catalog -- for indexed uncontrolled access points.
Besides being costly, redundant access can be confusing to users. For instance, what might a user make of the following:
In an online catalog a search under Tolkein, J.R.R. Lord of the Rings Two towers will retrieve:
A see reference:
Tolkein, J.R.R.
Lord of the rings, 2, Two Towers
see Tolkein, J.R.R.
Two Towers
AND records like:
100 Tolkein, J.R.R.
245 Two Towers
440 Lord of the Rings
100 Tolkein, J.R.R.
245 Lord of the Rings. $p Two towers
246 Two towers
Perhaps of even more concern is what systems designers might make of records like these.
In conclusion, work is needed on conceptual foundations as these relate to added entries and cross references. Redundancy results from not clearly delineating the roles of added entries and cross references. As illustrated above, the following problems attend such redundancy:
1. Only in a system without authority control is there a need for added entries for each member of a set of dependent titles or a set of dependent names.
2. Only in a system that is not automated is there a need for uncontrolled added entries.
3. In an automated system it would be useful to when retrieving items by added entries that are edition specific (Pope's translation of the Iliad) to automatically generate a see-also reference to the full work set (i.e. all editions of the Iliad).
4. Only in a system without authority control is there need to provide double access for a person who functions in more than one role.
Added Entries and Cross References
Authority work entails the making of cross references. It can be asked then, when should cross references be made? An easy answer to this question -- it is a policy question --is that cross references should be made to linking the variant names of an entity, e.g. the variant names of a person, a corporate body, a series or a work. There are, however, problems with this, theoretical collocation problems and practical problems in the AACR implementation of this policy.
The theoretical problem turns on what is meant by collocation. The second objective of the catalog require the collocation of the editions of a work and the works of an author be co-located or brought together in display. It is to fulfill this objective that authority work is done. But while the making of cross references is necessary to fulfill the second objective, with the cataloging rules the way they are to day this is not sufficient. Added entries are needed as well. For instance, in bringing together all the works of Pope, cross references are need in the regularizing of his name and his works. But also added entries are needed for when say his authorship is of a secondary character, for instance when he is the translator of the Iliad.
In Panizzi's day, added entries were not used; one "main entry" was made for the work being described and additional access points to that work took the form of cross references. Thus, Homer's Iliad translated by Pope would be entered under Homer, with a uniform title for the Iliad and a cross reference would be made from Pope, see Homer. Iliad. Today an added entry would be made for the item entered under Homer for Pope. What is the difference? The difference is that today, under the added entry system, a user searching under Pope would find only one edition of the Iliad; in Panizzi's time, when only cross references were used, he would find in the same search all editions of the Iliad collocated. When unit cards became available and added entries began replacing cross references at the end of the last century, the cost savings were obvious; the implications for collocation were not. Added entries and cross references were perceived in some sense as performing the same function. Thus, the 1908 rules state:
It is the opinion of Lubetzky, that this substitution of added entries for cross references was the first instance of many instances where ignoring the implications of a technological advance resulted in the dimming of the great vision of Panizzi, Jewett and Cutter..
The argument can be made, why if all the editions of the Iliad are collocated under Homer, is it necessary to lead the user to this collocation by additional access points, such as the translators of the Iliad. The answer simply is that this leading of the user to what he might want has been in the past a kind of moral imperative. Panizzi wrote that the user might know the work he wants, but he does not know how this work might be represented in the catalog, and this he has a right to know.
Lubetzky, probably history's greater champion for the cataloging imperative would argue for the eliminating of added entries. It is theoretically possible that a return to Panizzi and better collocation might be accomplished with present day technology. A proposal elaborating this will be made, but first a look at the difference between added entries and references is needed.
The function of added entries is to provide secondary access, while that of references is to provide equivalent access. Added entries are made to enable a user to find a work by an access point other than the one chosen as the main entry. The 1949 Rules state: the chief function of the added entry is to enable the user of the catalog to find a work when incomplete knowledge or imperfect memory of the work or unfamiliarity with the rules of entry would prevent ready access to the main entry.
References, on the other hand, are made to enable a user to find a work by forms of name other than the one chosen as authoritative. The chief use of references -- particularly see references, which are the kind under consideration in this paper -- is to map the user's expression to the heading used by the catalog for instance when the user enters a variant name of the author or a variant title of the work. In short, the function of added entries is to provide secondary access to a work, while that of references is to provide equivalent access.
A logical difference between added entries and references is that the former are, for the most part, attributes of works and the latter are not. The former represent attributes of the work being described in a bibliographic entry -- e.g. author or title. The latter represent equivalent relationships between alternatives names of such attributes. Another way of saying this is that the added entry is a variable whose values are the name of attributes of bibliographic works; whereas a reference is an equivalence relationships between two such values. Somewhat simplistically it might be said that added entries contribute primarily to the finding function of the catalog and references to the collocating function.
The current rules sometimes call for added entries when -- from a logical point of view -- cross references would be more appropriate. For example, the rule 25.2E which requires an added entry undertitle proper when the work represented by the item in had is entered under a uniform title. While making such an entry contributes to the finding function of the catalog, it has two disadvantages: (1) under the title proper added entry will be collocated only those manifestations of the work that happen to bear that title proper; thus, the collocating function of the catalog will not be met. and (2) situations can arise where a given title proper functions both as a "see from" title -- by definition, an unused heading -- and as an added entry -- by definition a used heading.
To insist upon discrete functions for added entries and references as the means of meeting the second objective of the catalog and preventing logical category mistakes would seem to entail not making added entries for titles proper when the works in question have explicit (i.e. 240 or 130) uniform titles. Title proper added entries would be made only when the title proper is the de facto uniform title.
To elaborate further upon the consequences of assigning discrete roles to added entries and references requires some clarification of terminology. In what follows "main entry" will be used to mean the citation form for a work consisting of a normalized name of the author together with the normalized title. If the work is entered under title, the main entry consists of the normalized title alone. Further, in what follows a distinction will be observed between a bibliographic entry and a reference entry, the former referring to a bibliographic record and the latter to an authority record.
An integrated catalog display consists of bibliographic entries and reference entries interfiled. An integrated display that meets the second objective will under a given search key display in the form of bibliographic or reference entries all names and titles associated with that key. For instance, suppose a search is for a person who is the primary author for some works and the secondary author for others. Displayed under the author's name will be:
Note that in the above only the first entry is a biblographic entry. The succeeding ones are reference entries. A display such as the above substitutes reference entries for bibliographic entries -- or another way of viewing this to say "added entries being made of work records rather than of item records. An instantiated example of how such a display would look is given in Figure 1.
The approach that would produce the above display would have three advantages; (1) it would be understandable; (2) it would satisfy both objectives of the catalog and (3) it would prevent internal logical inconsistency. The disadvantage that would appear to be serious is the considerable cost that might be incurred by the making of what would seem to be a great number of references. However, for many works, no references are needed beyond those need to link alternative forms of the author's name. The proposal assumes a uniform title for every work, at first sight a costly undertaking. However, It can be stipulated that where the uniform title for a work is the same as the title proper, the bibliographic entry may be substituted for the reference entry. In essence this is what is done presently.
There is a certain cost incurred already in designating author and title elements in added or secondary entries. From the cataloger view the situation before and after adopting the proposed rationalization would be the same: first access points must be determined and then authority work provided to normalize them. The rest is up to the systems designer how that information is to be used in display. It is true that in the proposed scheme many references are required. References are required from every author, every title and every author-title combination associated with a work to the author-title combination that represents the main entry for the work. However, it should be possible, theoretically at least, to have the computer construct the references with the information given it by the cataloger.
Two questions remain to be addressed. The first is what would be the effect on the cataloging rules of the proposed scheme. The second is what assumptions underlie it. Beginning with the second, it is assumed (1) that both objectives of the catalog are equally important; (2) that the collocation of the manifestations of works is part of the display function of the catalog; and (3) that added entries are properly not the attributes of items but of works. (This latter has serious consequences for the definition of work, e.g. earlier and revised tests would not represent the same work, as is the case now; but neither would translations -- Is this true? Work it out?)
Some thoughts:
The RI's for the 246 seem intent on eschewing authority work by making uncontrolled titles access points. On p. 14 is stated "...the function of providing access through a varying form of title is separate and disctinct from the function of collocation provided through a uniform title." Response: yes, indeed, the one provides access to a subset of the work, the other to the work. But access through any varying form of title on the piece, in an online catalog, can be achieved by keyword searching descriptive matter. Descriptive matter, including series, as it appears on the piece should be consolidated.
Editions of works have attributes. Some are also attributes of the work manifested by the edition. Most attributes are independent, e.g. number of pages, author, etc.; however some are not:
The chief function of the added entry is to enable the user of the catalog to find a work when incomplete knowledge or imperfect memory of the work or unfamiliarity with the rules of entry would prevent ready access to the main entry. [1949 Rules]
REDUNDANCIES CAUSED BY THE PRESENCE OF NONINDEPENDENT ATTRIBUTES
Return to Proceedings Home Page
| Advanced Search | Careers at OCLC | Feedback | Privacy Policy | ISO 9001 Certificate | ©2003 OCLC Online Computer Library Center, Inc. |