Authority Control in the 21st Century: An Invitational Conference
Authority Control in the 21st Century: An Invitational Conference
If you have examined your program, you will notice a title change. The original title sounded like I was going to tell a story. Once I started preparing for today, a better title became more obvious.
New catalog (INNOPAC) - July 1994 (part of statewide OhioLINK)
New method of authority file updating (BNA) - 1994-
Establishment of procedures
Staffing needs
PCC statistics - looked at most recent 3 years as indicative of current activity.
OSUL notifications - used the NOV 1994 notification records because they represented the largest group received and also, since the service was new, were a group for which we kept track of the bib. updates as part of developing procedures; also have additional records not examined for this presentation.
Looked at types of records changed, fields changed and some examination of the nature of the change.
Changes have different implications depending on one's point of view. There are maintenance issues (for the authority file and the bib. records), as well as issues for use of the catalog and the authority file.
| NEW n | % | CHA n | % | |
|---|---|---|---|---|
| NACO | ||||
| FY 93 | 74,763 | 83 | 14,895 | 17 |
| FY 94 | 81,302 | 82 | 17,770 | 18 |
| FY 95 | 96,655 | 80 | 23,794 | 20 |
| OSUL | ||||
| FY 94 | 786 | 88 | 107 | 12 |
| FY 95 | 730 | 83 | 150 | 17 |
First table shows the activity at the national level for contribution of new and changed authority records for names and series. LC reported these separately, but I grouped them because series are received as part of the BNA name notification service.
Note that while the level of contribution overall is growing, there is a trend toward increasing change. This was true of our own institution as well.
However, it was not true of everyone, as the next table shows.
| NEW n | % | CHA n | % | |
|---|---|---|---|---|
| LC CO | ||||
| FY 94 | 38 | 3 | 1028 | 97 |
| FY 95 | 106 | 7 | 1390 | 93 |
| UofVa | ||||
| FY 94 | 417 | 75 | 138 | 25 |
| FY 95 | 916 | 70 | 402 | 30 |
For the LC Cooperative Cataloging staff, there is a definite difference, in contributing more changed records than new ones. There was a slight decline in the rate of change.
I arbitrarily picked out one other NACO participant whose activity was somewhat different. In the case of the University of Virginia, there was a larger overall percentage of contribution of changed records, but they also showed a trend toward an increasing percentage of changed vs. new records.
Some of this, perhaps much, could be due to the increasing activity in series authorities, which tend to be more dynamic.
| NEW n | % | CHA n | % | |
|---|---|---|---|---|
| FY 93 | 1,031 | 84 | 198 | 16 |
| FY 94 | 918 | 81 | 214 | 19 |
| FY 95 | 1954 | 72 | 745 | 28 |
| day 91 | 19 | 38 | 30 | 60 |
Again, with subject authority records, the pattern is similar, with a greater jump this past FY in changes.
The number of name changes reported by PCC is about 1,982/mo.
OSUL received an average of 1,233/mo. for 9 of those months.
The first 3 months of LC's FY are excluded because they represent special notification from initial and gap loads rather than ongoing monthly notifications.
The number of subject changes reported by PCC is about 62/mo.
This is very odd in that OSUL received about 556 subject notifications/mo. I have not determined an explanation as yet.
| NAME = | NACO | FY 95 | 23,794 | 1,982/mo |
| BNA notif | JA-SE 95 | 11,096 | 1,233/mo | |
| SUBJ = | NACO | FY 95 | 745 | 62/mo |
| BNA notif | JA-SE 95 | 5,008 | 556/mo |
Calhoun & Oskins exam. of one mo. 1991 data of changes to LC AFs = N 60% new, 38% changed (not 80/20); S 38% new, 60% changed which was very different from the current trend.
As part of the development of a state-wide system, OhioLINK, OSUL migrated from locally-developed system, LCS, to OhioLINK Innovative Interfaces system.
Following a contest, renamed the local system OSCAR. Early this year, a duplicate of OSCAR, named FELIX, was created for staff functions primarily in automation and technical services.
The initial load of bib. and auth. recs. took place in Spring 1994, followed by gap loads. We have just received back from BNA bib. and auth. recs. for cataloging done Sept. 1994 to Dec. 1995, and these have not yet been loaded.
The initial and gap groups of bib. records excluded some materials: e.g., ERIC docs., OSU theses, and short records used for circulation of items without full cataloging in m-r form. The group included CRL loaded locally, as well as OSU Law Library and Ohio Historical Society.
ARs received from the bib. processing excluded those without cross-references or significant notes. These factors are relevant when looking at the ratio of bib. recs. to ARs, which results in about 17 ARs for every 100 bib. recs.
| NAME | SUBJ | MeSH | |
|---|---|---|---|
| Initial '94 | 304,900 | 117,371 | 5,546 |
| FY 94/95 | 97,401 | 21,865 | 541 |
| FY 95/96 | n/a | n/a | n/a |
| NEW = | 402,301 | 139,236 | 6,087 |
| FY 94/95 | 16,832 | 11,451 | n/a |
| FY 95/96 | 13,303 | 5,076 | n/a |
| NOTIF = | 30,135 | 16,527 | |
| #BIBRs | #ARs | AR/BIB | |
| March 22, 1996 | 3,214,218 | 552,074 | 17% |
| Total ARs | Name | 402,301 |
| Subject | 139,236 | |
| MeSH | 6,087 | |
| Total | 547,624 | |
| Deletes | -10,600 | |
| Total | ca. 537,024 |
Notif added rather than overlay = higher #
Some are correct adds
Some dups. not yet deleted
[chart of AR Loads FY 95/96]
| Typical mo. | NAMES: | range 801 for JAN 1996 changes |
| to 1778 for SEP 1995 changes | ||
| SUBJS: | range 337 for JAN 1996 changes | |
| to 766 for SEP 1995 changes |
JAN appears to suffer from post-holiday slump, while SEP is a month of high energy and perhaps fuller staffing following summer vacations.
Note that for these notification loads, about 72% are NAMES (inc. SER) and 28% are SUBJ.
| NAME ARs | SUBJECT ARs | ||
|---|---|---|---|
| 100 Pers | 1543 | 100 Pers | 337 |
| 110 Corp | 597 | 110 Corp | 184 |
| 111 Conf | 78 | 111 Conf | 8 |
| 130/1xx|t Series | 226 | 130/1xx|t Series | 1 |
| 130/1xx|t UniTi | 83 | 130/1xx|t UniTi | 28 |
| 151 Geog | 26 | 151 Geog | 58 |
| 150 Topic | 0 | 150 Topic | 1073 |
| TOTAL | 2553 | TOTAL | 1689 |
Looking at the sample month examined in detail, NOV 1994, here is the breakdown by the 1xx tag.
For this paper, all 2,553 NAME ARs were examined. For SUBJ ARs, only topical headings tagged 150 were included, a total of 1,073.
One tangential finding came from looking at the characteristics of the mr LCSH described by Vizine-Goetz & Markey in 1989.
| Tag of estab. hdg. | Total occ. | % of records | % of Nov 94 notifs |
|---|---|---|---|
| 100, 110, 111, 130 | 12,284 | 7.75 | 33 |
| 150 | 128,367 | 80.29 | 64 |
| 151 | 19,218 | 12.02 | 3 |
| TOTAL | 158,869 | 100.06 |
This table shows data about the 1987 mr LCSH, published in an article by Vizine-Goetz & Markey in 1989.
Not surprisingly, the largest number of records is for topical headings, tagged 150.
The last column shows that although sh tagged 100-130 represented only about 8% of the file, changed records are about 33% of those received in NOV 1994, and 64% of changed records were for topical headings, which represented about 80% of the file in 1987.
Geographic names apparently are fairly stable, at least for the time covered by the NOV 1994 notifications.
| TAG: | 100 | 110 | 111 | 130s 1xx|t | 130u 1xx|t | 151 |
|---|---|---|---|---|---|---|
| DLC | 65 | 64 | 55 | 87 | 93 | 81 |
| NACO | 35 | 36 | 45 | 13 | 7 | 19 |
| XXXX | XXX | XXX | XXX | XXX | XXX | XXX |
| 150 | n/a | |d | DLC |
Looking at the NOV 1994 notification records from BNA, this table shows whether the source of change was the Library of Congress or a NACO participant by examining the last |d in field 040.
Personal and corporate names show a similar pattern, with about 1/3 of changes by NACO participating libraries and 2/3 by LC. Conferences and meeting name changes, small in number, were made by NACO members in 45% of the cases.
For series ARs, changes are made primarily by LC, although this ratio could change with increasing participation by NACO members in series authority contribution.
Uni. titles seem to be avoided by NACO participants, at least in terms of changes to current records, as 93% of changes were made by LC.
Geographic name changes also show a pattern different from the other areas, with 81% by LC and 19% by NACO members.
The last |d in all SUBJ ARs is DLC, so this kind of comparison was not possible for that group.
| c/a/d | 100 | 110 | 111 | 130s | 130u | 151 |
|---|---|---|---|---|---|---|
| Pers | Corp | Conf | 1xx|t | 1xx|t | Geog | |
| 1xx | 8 | 7 | 17 | 12 | 12 | 0 |
| 4xx | 60 | 62 | 54 | 52 | 81 | 73 |
| 5xx | 4 | 37 | 35 | 20 | 2 | 12 |
| 642 | 0 | 0 | 0 | 7 | 0 | 0 |
| 663 | 0 | 0 | 0 | 0 | 0 | 0 |
| 667 | 1 | 6 | 0 | 10 | 4 | 12 |
| 670 | 79 | 66 | 54 | 37 | 45 | 77 |
| 680 | 0 | 0 | 0 | 0 | 0 | 0 |
| unkn | 9 | 4 | 4 | 10 | 4 | 8 |
| (c/a/d = change/add/delete) |
Each notification record was received in printout form, as well as on tape. The BNA report includes phrases indicating which fields were changed, e.g., 1xx changed, or 4xx added. These appear in a field tagged 699 at the end of each record. Each change is counted only once, so one 699 note for 4xx added could mean that 1 or more 4xx fields were added. These figures show what percentage of each type of record had at least one change to a field with that tag.
Some records lacked 699 fields. These Changes were counted as "unknown". Possibly these were 053 additions, or another change.
Most changes for NAME records were the addition of 4xx's, or see from cross-references. This was especially true of uniform titles, with 81% of records reviewing showing the addition of 1 or more 4xx fields.
This analysis also showed that for personal name records, 79% had a field 670 added. In many cases, this was the ONLY change, and usually contained what one might call biographical information - b. or d. date, fuller forms of names, or other identifying information not needed to estab. the hdg. but to serve a potential reference need.
| c/a/d | 150 Topic |
|---|---|
| 1xx | 4 |
| 4xx | 23 |
| 5xx | 16 |
| 670 | 16 |
| 680 | 1 |
| unknown | 63 |
One obvious difference between the name and subject notifications is the higher percentage of unknown changes, with names ranging from 4 - 10 % for the different types of records. I believe the high percentage of unknown changes to subj. notifications is the addition of geographic subdivision.
I checked several of the many hdgs. in this report in printed volumes of LCSH beginning with the letter "p" and this was the only identifiable change to the authority record. Examples include:
As for NAMES, most of the 4xx changes were the addition of a 4xx field. Unlike NAME records, however, the 5xx changes were about evenly divided among changes (5.5%), adds (4.5%), and deletes (5.5%).
[this detailed breakdown is available on spreadsheets]
| 1995 | TOTAL n | 1xx n | 1xx % |
|---|---|---|---|
| (Records) | (Changed) | ||
| Jul Name | 1,670 | 299 | 17.90 |
| Jul Subj | 560 | 64 | 11.40 |
| Aug Name | 1,291 | 161 | 12.50 |
| Aug Subj | 379 | 39 | 10.30 |
| Sep Name | 1,778 | 213 | 11.97 |
| Sep Subj | 766 | 89 | 11.60 |
| Oct Name | 1,320 | 182 | 13.80 |
| Oct Subj | 490 | 68 | 13.90 |
This table shows the pattern of percentage of changes to 1xx headings.
The figures are taken from record counts on disks of 1xx changed records provided by BNA, and the record count of the full files received on tape.
I bring this up for comparison to NOV 1994, where the range of percentages for changes to 1xx hdgs. went from a low of 7 % for corp. names to a high of 17% for conference or meeting names.
For subjects for NOV 1994, the 1xx changes reported were only for 4% of records examined, but this group included only records tagged 150.
The inclusion in this chart of all subj records, including those tagged 100-130 and 151, may explain the large difference in percentages.
These figures for name changes are comparable to that found by Calhoun & Oskins in a 1992 report of Rates and types of changes to LC authority files, where data analysis from 1 month in 1991 showed that slightly more than 1 change in 10 is a change to 1xx hdg.
The two areas of maintenance which have required most of our attention since we began notification loads are de-duping and updating of existing bib. records when a 1xx changed after BNA authorities processing.
Duplicates resulted when, during what is called “overlay”processing, the notification record does not replace the existing record but is added instead. Occurs when there are, in a small number of cases, sequential updates on the same tape. If “target record is busy” being replace by the first notification, the second and later version is added. This past month, we began receiving files of notification records which had been de-duped by BNA so that we only have the latest version for overlay.
Duplicates also result when the same AR is present in both NAME and SUBJ indexes, and the system did not know which one to overlay, and so added the update as a new record. We have to date deleted more than 10,000 duplicate records. This problem is expected to be fixed with the next Innovative Interfaces release, coming soon to a catalog near you.
| Updates: | # of hdgs. | # of bibs. |
|---|---|---|
| 1994 NOV | ||
| Global | 142 | 136 |
| FSE | n/a | 141 |
| 1994 DEC | ||
| Global | 144 | 139 |
| FSE | n/a | 54 |
| 1995 Jan | ||
| Global | 137 | 132 |
| FSE | n/a | 53 |
This is a sample of local update transactions based on 1xx changed hdgs. from NOV and DEC 1994 and JAN 1995. There does seem to be a pattern emerging.
Global updates are used for both NAMES and SUBJS. Likewise, single updates to bib. records also are used in both cases.
The effort required for each 1xx change depends on the situation for each hdg. Variants must be identified, there are often language difficulties, and subj. hdg. rotation in INNOPAC makes it more difficult to identify occurrences of a hdg.
Here are just two examples of NAME changes. The corporate name change reflects rule changes, the personal name change is probably the result of a marriage and usage of a later name, hyphenated name.
I did not attempt to compile a complete list, so these are not given as typical, but I suspect they are not uncommon.
These changes to subject headings likewise are not surprising.
One indicates a trend toward fewer inverted hdgs.
The value of the last change is unclear to me, but the change in terminology presumably is meaningful in the business vocabulary.
The changes to established headings represented a small number of all changes to ARs examined. Most changes were to 4xx, 5xx and 670 fields.
Here are a few examples of the types of 4xx additions.
The first name is for a living individual, who used his middle name very publicly. With the change to this record, he also can be found by his real first name, Andrew.
The second name is for an individual who is gone but not forgotten. In this case, a reference was added from a form by which he was commonly known. There were other cases of names for persons gone a thousand years, but they tended to be very long Arabic names involving diacritics so I passed up on those to fit things on one slide and avoid characters not part of the PowerPoint set.
The last case again arises from usage. Mr. Macpherson sometimes referred to himself by his nickname, which was added as a cross-reference.
Two examples from uniform title changes present very different cases.
In the example for the Bible, the reference is for a specific title in another language.
The case for the musical work is typical of these uniform titles, bringing works together under Selections. This also popped up several times in the group examined for collected works of cartoonists published in book form.
These additions of see from references to subjects illustrate two types of changes. The plural form, cakes, to cake, helps a user who may approach the catalog with that term in mind.
One example of this type that comes to mind was brought to my attention recently by one of our reference librarians, trying to help a patron locate material about turkeys, but who could only find items about the country. I explained the principle of plurality about things, which does not completely hold up where these two edible items are concerned. In case you were wondering, Pies is plural, but Pastry is singular. And no, there are no cross-references from Pie or Pastries.
In scanning for examples, there were many instances of the second type, adding 4xx’s for vocabulary from other domains. Many of the records were for business topics, so there was a lot of activity in that area in that particular month.
I mentioned earlier that for personal names, about 80% of the records had 670 fields added. Some of these 670 justified the addition of 4xx or 5xx fields, but for many records, the 670 addition was the only change.
The 670 provided in all cases additional information about the person, in the form of dates or other identifying information. Often, when the change was made by LC, the information came from the CIP data sheet, a source which is discarded.
The value of the 670 information is that it forms part of an online reference tool, which is readily accessible during the cataloging process. One does not have to trot off to reference shelves somewhere, or connect to a different online source.
Most of you are aware of the reasons for additions of 5xx references to ARs.
This just highlights the obvious reasons, and to save space and time, I just presented them with record numbers in parens.
As the data analysis showed earlier, only about 4% of personal name records had a change or addition to a 5xx field, with 37% and 35% for corp. and conf. names respectively, and 20% for series. For geog. names, the percentage of records with this change was 12% . And for subjects, 16% of the records had a changed or added 5xx.
At the beginning of this presention, I mentioned three areas of implications for changes to ARs: maintenance of the local catalog catalog use, and authority file use.
I have described a bit about how the load of notification records has resulted in maintenance for our local system, in terms of the de-duping (for now), and then subsequent updating of the existing bib. recs.
For catalog use, 1xx changes mean, generally, that currency of the access point is improved. Cataloging rules and common usage are reflected in the changes, and changes such as subj. simplification.
The added cross references serve as added entry vocabulary and navigation aids, both to public and staff users. Hildreth (1982) said that cross-refs. facilitate term selection browsing and Rice said that when there are more access points, serendipity is less accidental and more systematic. In my own 1988 study, 14% of users seaches matched a sub. xref.
5xx refs. are of course a navigational aid to both public and staff users.
Cataloging practioners, especially authorities coordinators, know the value of the elements of the authority file which I have talked about today - the 4xx, 5xx and 670 fields.
Without these, an authority record is skeletal, which may be adequate for many names and subjects if the established hdg. is sufficient in the broader bibliographic context.
But the cross-references of the AF serve as added access points and navigational aids to us as well, enabling us to expedite cataloging when a heading is established and the record can be found quickly.
The value of the 670 was mentioned earlier. As an online reference tool, the AF is highly accessible (not to mention free) and is as current as we choose to make it.
I have described the characteristics of change from one view, to show a sample of the volume and types of changes received on BNAs notification service
Another, broader view of the characteristics of change to ARs is that it is inevitable, predictable, and beneficial.
Usage is a driving force. Cultural changes, personal preferences and inconsistencies in name usage, corp. changes, vocabulary changes all can affect the AF. Cataloging rules and to some extent cataloger judgement are another element driving change. The AF reflects and documents rules and standards, both AACR and LCSH, and is being extended to include international views.
At issue is are we making the right changes or enough changes.
Is more better? more of the same? or something different or additional?
Finally, it is worth remembering that changes to the AF do not sit in isolation, but are reflected in whole or part in local catalogs. When ARs are enhanced, in many cases the direct result is improved access in local systems as well.
Return to Proceedings Home Page
Advanced Search | Careers at OCLC | Feedback | Privacy Policy | ISO 9001 Certificate | ©2003 OCLC Online Computer Library Center, Inc. |