Standard for Machine-Proposed Authority Records
Gary L. Strawn
Northwestern University Library
mrsmith@nwu.edu
Page 1 of 3.
NOTE: Every attempt has been made to include the formatting present in the
source document in this version. However, some features, such as spacing, indentions
and diacritics cannot easily be reproduced here. Contact the author to get
a copy of the document in Microsoft Word format.
1. Introduction
An increasing number of automated library systems can create authority records
from headings in bibliographic records. The contents of these machine-produced
authority records vary, depending on the circumstances under which the authority
records are created and the capabilities of the system. An automated library
system which requires an authority record for every unique heading may create
skeletal records (consisting of little more than the heading itself) as the
system encounters new headings; other automated library systems may allow an
operator to request authority records (typically containing more than just
the heading) at any time.
This standard describes the contents of one type of authority record produced
by computer program: those authority records which are individually requested
by an operator and are intended for eventual contribution to a national-level
cooperative authority file as new records. These authority records are called
machine-proposed authority records, because they are prepared by a program
and presented to an operator for inspection; the operator bears final responsibility
for the record. Machine-proposed authority records are to be distinguished
from machine-generated authority records, which are created by a program and
entered directly into a local authority file without operator review.
Authority records produced in accordance with this standard are intended to
meet the needs of those working with a variety of headings in a variety of
languages found in bibliographic records for materials in a variety of formats. [See
Endnote 1] Because not all of the features defined in this standard
may be appropriate in every situation, a program which offers all of these
features may allow an individual operator to control the behavior of the program.
The provision of user options can allow a full-fledged authority record creation
program to be adapted to a variety of unusual or special circumstances.
The machine-proposed authority records described in this standard are produced
exclusively by the manipulation of a heading and other information extracted
from a bibliographic record, by the drawing of inferences from that information,
and by the judicious use of default values. Because the information needed
for an authority record is not necessarily all present in a recognizable form
in the bibliographic records which carry the heading to be established [See
Endnote 2], and because computer systems work from a set of rules
which can at best only approximate correct behavior for the majority of cases
and cannot be expected to handle every possible exception [See Endnote
3], the operator asking a program to propose an authority record must
evaluate each element of the record, and must adjust the record as necessary
before either adding the record to the local authority file, or forwarding
the record to a national-level cooperative authority file. [See Endnote
4]
Although this standard is written as if authority records will be created
on a local automated library system, and later contributed to a national-level
authority file, this standard may also be applied to programs which create
authority records directly on the system which maintains a national-level authority
file. This standard does not define the manner in which a locally-created record
residing on a local system may need to be manipulated to bring it into a form
suitable for such contribution. This standard also does not define the manner
in which the information present in a record transmitted to a system maintaining
a national-level cooperative authority file should be handled by that system. [See
Endnote 5]
2. Definitions
This standard assumes a general knowledge of library terminology. The following
terms have specialized definitions in this standard.
- alphanumeric character
- One of the following characters: A-Z, a-z, 0-9.
- authority heading
- The 1XX field in an authority record.
- conference heading
- A heading tagged X11 in the source bibliographic record.
- forename
- If the first indicator of a personal name heading is "1" or "2" and if
subfield $a contains an internal comma, the forename is that portion of subfield
$a following the first internal comma; if subfield $a does not contain an
internal comma, the name has no forename. If the first indicator of a personal
name heading is "0," the forename is the entire subfield $a.
Examples:
100 1 $a Boop, Alexander von der, $d 1952-
Surname: Boop
Forename: Alexander von der
100 2 $a Schleswig-Holstein-Boop, Friedrich Christian von, $d 1952-
Surname: Schleswig-Holstein-Boop
Forename: Friedrich Christian von
100 1 $a De La Boop, $d 1952-
Surname: De La Boop
Forename: none
100 0 $a Mobutu Sese Boop, $d 1952-
Surname: none
Forename: Mobutu Sese Boop
- heading
- An access point in a bibliographic record which is subject to authority
control: the 1XX, 400-440, 600-651, 655, 700-730 and 800-830 fields. A heading
may simultaneously fit into more than one of the categories recognized here;
for example, a heading may be both a personal name heading and a topical
heading [See Endnote 6], or a conference heading and a series. [See
Endnote 7]
- name/title heading
- A heading tagged X00, X10 or X11 which contains subfield $t; subfield $t
and following subfields (up to any subfield $v, $x, $y or $z) constitute
the heading's title segment. A bibliographic 100-111 field may also be combined
with the bibliographic 240 field (or, in rare cases, the bibliographic 245
field) to form a name/title heading; in this case, the bibliographic 24X
field becomes the title segment.
- operator
- Any person using a program to create authority records.
- personal name heading
- A heading tagged X00 in a bibliographic record.
- program
- Any automated program, suite of programs, automated system feature or suite
of features that provides the ability to produce an authority record from
information in a bibliographic record on operator request.
- series
- A heading tagged 400-440 or 800-830 in a bibliographic record. However,
if the source heading is tagged 400-411 or 800-811 and the heading is not
a name/title heading (because less than the complete heading is being established),
the heading is not a series heading.
- surname
- If the first indicator of a personal name heading is "1" or "2" and if
subfield $a contains an internal comma, the surname is that portion of subfield
$a up to the first internal comma; if subfield $a does not contain an internal
comma, the surname is the entire subfield $a. If the first indicator of a
personal name heading is "0," the name has no surname.
See examples under the definition of forename.
- topical heading
- A heading which in a bibliographic record can only appear in a 6XX field.
In a bibliographic record, a topical heading is tagged 650 or 655; or is
any field tagged 600, 610, 611, 630 or 651 which contains subfield $v, $x,
$y or $z.
Examples of bibliographic topical headings:
650 2 $a Bacterial Proteins.
650 0 $a Hail control $x Congresses.
600 10 $a Shakespeare, William, $d 1564-1616 $x Homes and haunts.
651 0 $a United States $x History $y 20th century.
630 0 $a Bible. $p N.T. $p Ephesians $x Criticism, interpretation, etc.
Examples of bibliographic headings which are not topical headings:
600 10 $a Shakespeare, William, $d 1564-1616. $t Hamlet.
710 2 $a Radio Corporation of America.
440 0 $a Maseru development plan working paper
630 0 $a Chanson de Roland.
- unadorned heading
- A heading which does not contain subfields $t, $v, $x, $y or $z.
Examples of unadorned headings:
650 2 $a Bacterial Proteins.
100 1 $a Shakespeare, William, $d 1564-1616.
710 1 $a United States. $b Congress. $b House of Representatives.
- word
- One or more characters separated from adjacent words by spaces or hyphens.
3. Levels of completeness
This standard distinguishes three broad levels of completeness in machine-proposed
authority records, based on the amount of effort expended by the program in
the preparation of the record. These three levels do not represent absolute
criteria which must be met exactly, but instead define three base levels upon
which system designers can build additional features and enhancements in harmony
with the principles described in this standard. In order for a particular program
to be said to produce authority records which satisfy the requirements of a
particular level of this standard, its proposed authority records must contain
at least the degree of completeness specified for that level; but the program
may exceed the amount of detail specified for a particular level.
Regardless of the level of the initial authority record provided by a particular
program, the operator is expected to bring each record up to "full" level before
approving it for addition to the local authority file or contribution to a
national-level cooperative project; there is consequently no correspondence
between the three levels for machine-proposed authority records described here,
and an authority record's encoding level (Leader/17) or level of establishment
code (008/33).
The three levels of completeness for machine-proposed authority records recognized
by this standard are:
- minimal
- A minimal-level record consists of a Leader and a Directory, and the following
variable fields: 001, 003, 005, 008, 040, 1XX and 670, as specified below.
For series headings, the record also contains the applicable 022 and 642-646
fields, also as specified below. A program producing authority records at
the minimal level makes no attempt to supply reference tracings as part of
the proposed authority record.
- intermediate
- An intermediate-level record contains all of the fields defined for a minimal-level
record. In addition, the program shall include in the proposed authority
record at least some of the categories of reference tracings defined for
maximum-level records; these reference tracings may require substantial adjustment
on the part of the operator. (For example, a program creating authority records
at the intermediate level may only generate reference tracings for multi-word
personal names; as it does this, it may make no attempt either to eliminate
reference tracings not likely to be needed or to adjust the first indicator
or the contents of any subfield $q to suit the context of a particular reference
tracing.)
- maximum
- A maximum-level record contains every field which can reliably be supplied
through the machine manipulation of bibliographic information, including
all of the fields defined for intermediate-level records. A program creating
authority records at the maximum level is able to generate at least all of
the categories of reference tracings defined in this standard. Most importantly,
a program creating authority records at the maximum level makes a substantial
attempt to adjust the contents of the reference tracings and 670 field to
match as closely as possible the assumed final appearance of these fields;
the program reduces to the extent possible the amount of cleanup required
of the operator.
To state these categories in another way: Programs operating at the intermediate
and maximum levels are distinguished from programs operating at the minimal
level by their attempt to generate reference tracings. Programs operating at
the maximum level are distinguished from programs operating at the intermediate
level by the completeness of the suite of reference tracings they produce,
and in the construction of subfield $b in the 670 field.
4. Data elements
4.1 Leader
The following paragraphs describe the values the program shall supply for
each Leader position, in terms of the USMARC authority format. [See
Endnote 8] Elements described as "system-generated" are formulated
according to the USMARC format. Some automated library systems do not require
all of this information for internal use; these systems must supply the default
data elements when an authority record is transmitted beyond the local system.
Leader/00-04 (record length): system-generated
Leader/05 (record status): "n"
Leader/06 (type of record): "z"
Leader/07-09 (undefined): blanks
Leader/10 (indicator count): "2"
Leader/11 (subfield code count): "2"
Leader/12-16 (base address of data): system-generated
Leader/17 (encoding level): "n"
Leader/18-19 (undefined): blanks
Leader/20-23: "4500"
4.2 001, 003 and 005 fields
The program will formulate any 001, 003 and 005 fields (or their equivalents)
the local automated library system requires for its own purposes. If the local
system does not require one or more of these fields for its internal use, it
must supply them when a record is transmitted beyond the local system.
4.3 008 field
The following paragraphs describe the values the program shall supply for
each 008 field position, in terms of the USMARC authority format. Elements
described as "system-generated" are formulated according to the USMARC format.
Some automated library systems support the use of additional codes in the 008
field; these codes may be supplied for local use instead of those indicated
here. Locally-defined 008 codes must be translated into their USMARC equivalents
when an authority record is contributed to a national-level cooperative project.
008/00-05 (date entered on file): system-generated
008/06 (direct/indirect geographic subdivision): blank for topical headings; "n" for
other headings
008/07 (romanization scheme): fill character
008/08 (undefined): blank
008/09 (kind of record): "a"
008/10 (descriptive cataloging rules): for non-topical headings: derived from
the descriptive cataloging code (Leader/18) of the source bibliographic record
("c" if bibliographic Leader/18 is "a"; "a" in other cases); for topical headings: "n"
008/11 (subject heading system): for headings derived from bibliographic 6XX
fields, a translation of the bibliographic field's second indicator [See
Endnote 9]; for headings derived from other bibliographic fields,
the local institution's default subject system code.
008/12 (series type): for series headings: "a" for series unless entered under
personal name, in which case "b"; for non-series headings: "n"
008/13 (series numbering): for series headings: "a" if subfield $v is present
in the bibliographic heading, "b" in other cases; for non-series headings: "n"
008/14 (main/added use): for topical headings: "b"; for other headings: "a"
008/15 (subject use): "a"
008/16 (series use): for series headings: "a"; for other headings: "b"
008/17 (subdivision type): "n"
008/18-27 (undefined): blank
008/28 (government agency): fill character
008/29 (reference evaluation): "a" if reference tracings are generated by the
program; "n" in other cases
008/30 (undefined): blank
008/31 (update in process): "a"
008/32 (undifferentiated personal name): for personal name headings: "a"; for
all other headings: "n"
008/33 (establishment level): "a"
008/34-37 (undefined): blank
008/38 (modified record): blank
008/39 (cataloging source): "d"
4.4 040 field
The program shall supply an 040 field consisting of the local institution's "NUC" code
in subfields $a and $c. (This requires that the program be aware of the local
institution's NUC code.) If 008/11 (subject heading system) is "z," the program
shall supply subfield $f between subfields $a and $c.
4.5 Heading (1XX field)
The program shall extract from the bibliographic record the portion of the
variable field or fields from which it is to create the authority heading. [See
Endnote 10] The program shall adjust the tag and indicators of the
extracted field as necessary to make them conform to the specifications for
authority 1XX fields. [See Endnote 11] The program shall remove
from the extracted field(s) those subfields which do not form part of the uniform
heading (subfields $u and $4, for example). The program shall remove non-filing
characters from the beginning of the extracted 1XX field. [See Endnote
12]
To construct an authority heading from the bibliographic 1XX+240 or 1XX+245
fields, the program shall add a full stop to the end of the bibliographic 1XX
field (unless this field ends with a hyphen or a full stop), and change the
initial $a code in the 24X field to $t. [See Endnote 13]
The program shall attempt to remove unnecessary terminal punctuation from
the authority 1XX field. [See Endnote 14]
4.6 Reference tracings (4XX fields)
4.6.1 General statements
An authority record production program at the maximum level shall attempt
to generate appropriate reference tracings for each new authority record at
the same time it creates the remainder of the proposed authority record.
These principles should guide the design of programs intended to generate
reference tracings at the maximum level:
- The program should attempt to generate all reference tracings which may
predictably be required to accompany a heading in nearly all cases, and should
avoid the generation of unnecessary or irrelevant reference tracings the
operator will have to delete in nearly all cases. [See Endnote 15]
- Reference tracings generated by the program should, to the extent possible,
not require additional operator manipulation; for example, a program at the
maximum level shall adjust the first indicator and the contents of subfield
$q in personal name headings to suit each new situation.
A program creating authority records at the intermediate level shall also
make an effort to generate reference tracings for the authority record when
initially proposed, although it may not necessarily be able to generate all
of the categories of reference tracings produced by a program at the maximum
level; in addition, reference tracings produced by a program at the intermediate
level will in some cases require more attention from the operator than the
same reference tracings produced by a program at the maximum level, before
the reference tracings are acceptable for use in an online catalog. A program
generating authority records at the minimum level will make no attempt to generate
reference tracings for the authority record when the record is initially proposed.
A program at the maximum level shall be able to generate at least the reference
tracings shown for the following model headings [See Endnote 16] and
shall not generate reference tracings for these headings identified as not
wanted.
The examples show the reference tracings as generated by a program operating
at the maximum level, before operator review.
Next page
|