Is Faster Better?
An Observation of Chinese Vendor Records
By
Shuyong Jiang
Abstract
In the past, observations of vendor records have been negative. Loading vendor
records to a library catalog in batches might be the fastest method, but library professionals
are concerned about these low-quality and non-standard cataloging practices. Recently,
several Chinese vendors began to beta test their MARC data services; when asked, they
provide MARC records along with the order lists. Libraries can load these records directly
to utilities such as OCLC, or into their local systems. Since these services haven’t been
widely adopted, many still question the quality and usability of these records. Vendor
records will make library material available quickly, but would this increased “processing
speed [jeopardize the] meticulous quality” of cataloging standards?1 This paper will
discuss the issues of quality control, effectiveness, and the economic impact of vendor
records by evaluating MARC records from a major Chinese vendor, as they are used for
acquiring and cataloging Chinese materials. In so doing, this paper will further explore the
implications of vendor records by investigating the effect this new process may have on
our assumptions of cataloging and technical services in general.
Keywords: Vendor records, cooperative cataloging, Chinese cataloging, MARC
1
Introduction
Recently, several Chinese vendors launched MARC data services, which provide
descriptive data of Chinese language monographs, among other relevant information.
Libraries can load these records directly to utilities such as OCLC, or into their local
systems. Each of these records includes parallel fields for both romanization and original
scripts. This format, if accurately standardized, will save a great deal of time for inputting
bilingual data in cataloging, especially for original scripts of Chinese, Japanese and
Korean. Currently, only a few libraries in North America are using vendor records for
acquisitions, and no libraries are using full MARC services even though they are available
from Chinese vendors. Since Chinese vendors started these services less than a year ago,
this indecision could result from insufficient assessment of, or a lack of confidence in,
vendor records.
Nevertheless, vendor records are not a new thing. As early as 1996, OCLC and
RLIN were already adding bibliographical records of European language monographs
provided by foreign book vendors. But concerns were raised about the low quality, and the
incomplete, non-standard practices that vendor records have displayed.2 While there are
several studies about vendor records of European language monographs, no studies have
yet been conducted to evaluate Chinese, or other East Asian languages, vendor records.
Are the concerns raised about European vendor records also true to Chinese vendor
records? What is the quality of Chinese vendor records? Are any particular problems
associated with Chinese vendor records? What kind of improvements might be required?
These questions crucially affect the implementation of vendor records into the technical
processing of Chinese materials; and thus this observation is structured by this interrelated
2
cluster of questions. As many libraries seek ways to improve cataloging and access, or
consider implementing vendor records as part of processing, it becomes necessary for them
to obtain relevant data about the quality and usability of vendor records. Based on the
evaluation of MARC records from a major Chinese vendor, this paper will discuss the
implications of vendor records in the light of some key elements of cataloging: efficiency,
timeliness, and cooperation. Hopefully, this limited study of Chinese vendor records will
be followed by more studies of East Asian vendor records, which will help produce a
better understanding of the current status of vendor records, and encourage libraries and
cataloging utilities such as OCLC to adopt high quality records for efficient cataloging of
East Asian materials.
Collecting data
The vendor records used in this study come from a major Chinese book vendor,
China International Book Trading Company (CIBTC). These records are brief MARC
records and come in .mrc, .iso, or .db format. They usually contain parallel fields of 1xx
and 7xx for names, 245 for titles, 4xx for series and 260 for publishing data. The
bibliographical data are automatically extracted from sources that are provided by
publishers. Although they are machine-generated records, the vendors did claim that all its
records are manually reviewed. This may sound promising, but we need to consider that,
following the typical workflow of vendor records, many of these records are created before
the books are published. This explains some of the problems found in these specific
records. While some of them can be easily fixed by reprogramming the process or
softwares, some are inherent and may inevitably affect the quality of vendor records.
3
Efficient cataloging means spending minimal time per record. Libraries in North
America are interested in implementing vendor records because we want to save time, but
minor changes could have an impact on the cataloging speed. Thus, when we evaluate
vendor records, we must consider how much work needs to be done for bibliographical and
authority control. In Shedenihelm and Burk’s study of Spanish vendor records, the first
question they asked was whether there were errors in vendor records that made them
difficult to utilize as acceptable records that meet the national cataloging rules and
standards.3 In this study, I have collected data to see how often and what type of errors
occur in order to provide a general evaluation of the cataloging quality.
One of the major errors associated with vendor records is the non-standardized
form of names used; and this lack of authorized headings in vendor records has the
potential to be very costly over time.4 To assess the use of authorized names, I have
compared the names in the sample records with the Library of Congress name authority
files in OCLC. In addition to the names, the lack of subject headings and classification
numbers make vendor records less usable, unless they are otherwise enhanced. Because of
these problems, some believed that it was easier and more cost-effective to create an
original cataloging record.5
In collecting data, I have also conducted searches to see how many of these records
will need original cataloging. I investigated the gap period between the time when an item
was published and when the first full bibliographic record in the OCLC was created. The
rate of original cataloging will affect a library’s decision to implement vendor records by
revealing the timeliness of cataloging. Once these data are collected and presented, I will
4
discuss the level of cataloging, core elements of records, and the various factors that affect
the quality of vendor records.
Collected Data
The order list used in this study was randomly picked from the orders recently
placed with the vendor to keep it in a normal working setting. A set of one hundred and
thirty two items are on the list with publishing dates ranging from January 2004 through
January 2006, except for four items that were published before 2004. During May, two
student assistants searched the OCLC database for all the records. Twenty-three out of one
hundred and thirty two did not match the records in OCLC, as show in the table 1.
Table 1
No Records Total
Has Records
109 + 5* 23 -1* 132 + 4*
# of records
83.82 16.18 100
% of records
*One of the titles was listed under series, while the OCLC records were cataloged individually each of the
five volumes.
The gap between the publication date and cataloging date can be as short as three months.
For example, an item was published in January 2006 and was cataloged in April 2006. All
those without an OCLC match record were published after 2005. The low rate of original
cataloging and short turnaround time suggest that in OCLC among the Chinese cataloging
community, records are created in a very timely fashion.
For descriptive data of the records, I only considered the one hundred and nine
records that match the OCLC entry, because without real items, it will be difficult to verify
5
the data in those records that do not have the OCLC match. I measured the completeness
and the accuracy of title information in the 245 field by using existing full OCLC records.
The results are listed in table 2 below.
Table 2 titles
Type of errors # of records % of 109
lack of a subtitle 6 5.5%
lack of a parallel title 17 15.6%
incorrect title (tran. spelling) 6 5.5%
improper data in title 10 9.18 %
incomplete title info 4 3.67 %
Total 43 39.44%
In addition to the title information, I also evaluated the choice of the main entry and the
accuracy of the description of responsibility areas, since these will have an impact on
access. I cross-referenced the forms of the names used with the LC Name authority files
and OCLC existing records in order to validate the accuracy of the relevant headings.
Table 3 Names
Type of errors # of records % of 109
Names without dates 35 32.11%
Wrong names 5 4.59%
6
incomplete responsibility 5 4.59%
incorrect choice of main entry 14 12.85%
Total 59 54.13%
From my research, it is evident that no authority is consulted for name headings in the
vendor records. Not only are names entered in multiple forms, or unauthorized forms, I
also found several examples where foreign names were entered as Chinese names.
31 records out of 109, or a 28.45% of total records, are error free in both title and
name fields. The other 78% of these are problematic. Many records have multiple errors.
If we suppose that each record has one error, the rate of error records is 93.47%. These
findings tend to agree with previous studies on vendor records for European languages; the
two most problematic areas are consistently: 1) choice and form of headings and 2)
description of the items.
Discussion
As the collected data has shown, Chinese vendor records seem to fall into the usual
pattern of vendor records, displaying characteristic types of errors. Some of these errors
may be avoidable with a better profile design for machine conversion, but some are
inherent to the way the vendor records are created and require human attention.
Because vendor records are machine generated from publisher provided sources,
some of them are pre-printed. Oftentimes, a vendor record is created before a real item is
available. This creates the possibility for incompleteness and errors that can potentially be
avoided. For example, there are four records for conference proceedings on the list used for
7
this study, all of them failed to include proper conference information in the title field and
main entry, because this type of information usually does not appear in a publisher’s
catalog. Likewise, all the parallel title information is also missing in these records because
of publisher catalog conventions. The physical descriptions of items are also unavailable.
Without real items at hand, these data are not possible to obtain. Still, although these
records tend to lack a variety of basic information, some unnecessary details are often
inexplicably included. For example, one may find volume information inserted in the title
proper (Figure 1). Series title information also appeared together with titles of individual
volumes in the title field, instead of the series title field (Figure 2). While some
information will not be possible to include without consulting the real item, some of these
unnecessary additions could be cleaned up.
Figure 1 Figure 2
Again we can see that no authority verification has been performed on these
records. There are 35 names that did not include dates, two of them are filed in the wrong
forms, and both are foreign names (Figure 3). This does not mean that the rest of the 74
8
records are valid names, it only means that these name headings happened to exclude
dates. In this regard, human effort can make some difference. The Library of Congress
authority files are available to cross-reference. According to the vendor, all the records are
reviewed after being automatically converted from its database. Training reviewers to
better search OCLC and LC databases would increase the reliability of name entries in
vendor records.
Figure 3
Other problems such as the incorrect spaces between sub-fields, and lack of
publishing locations can also be fixed and will make a big difference in terms of timeliness
and efficiency. Because non-Roman records have parallel fields, any type of error will
need to be fixed in both fields, and that will waste double the time per record. For example,
to delete the unnecessary volume information in title proper, as shown in previous
example, the same deletion would need to be done twice. Such redundant efforts would
significantly decrease the timeliness and efficiency of cataloging.
9
There is no question that the quality of Chinese vendor records needs to be
improved to become usable in standardized bibliographical databases. However, there are
some new possibilities that would benefit effective cataloging. Chinese vendor records can
be loaded into local system instead to directly loading into OCLC. Librarians can thus
enhance and modify the records before they are added into the OCLC to eliminate the
problems and complaints about the vendor records imported to those databases. This
involves some workflow changes and cooperative efforts. The concept of cooperative
cataloging could reduce the amount of duplicated labor and effort at the individual, local,
regional, and national level.
Conclusion
This brings us back to the basic question: Can vendor records have an impact on
cataloging speed, and in what way, positive or negative? With regards to the unique nature
and needs of acquiring and processing non-Roman materials, vendor records have the
potential to help us streamline the acquisition procedure and cope with the high demand of
processing.
In the Asian Library at the University of Illinois at Urbana-Champaign, true to
many other East Asian libraries, the process of acquisition has many repeated steps. After
selecting vendor records to order, we, the librarians, have to type Chinese characters for
the order list, then we will have to do multiple pre-order searches in our order file, the local
system and even in the card catalog to check for duplications. With our newly upgraded
local system (UIUC uses the Voyager 6.1 with Unicode enhancement), the Chinese
character search function is now in effect, so we can spend one-third of the time on pre-
10
order searches. Furthermore, because all the ordered items will be in the local system, the
new procedure of loading vendor records into the local system when the order is placed
can also better prevent duplicates.
Using vendor records for acquisition will also allow the ordered title to appear
immediately in local database at the time the order is placed. The timeliness is another
benefit of implementing vendor records. Many East Asian libraries have been struggling to
keep the backlog down and to make un-cataloged items accessible to users. Vendor records
can certainly help to make that happen. When we talk about the timeliness of cataloging,
we are ultimately concerned with how soon bibliographical information of ordered items
will be made available to users. Those who have a huge backlog would probably agree,
and some have already made progress by implementing the use of vendor records.
Implementing vendor records is also cost effective and time saving, at least when it
comes to inputting parallel fields of Chinese original scripts. In cataloging Chinese or other
non-Roman materials, data needs to be entered twice for most of the fields. When inputting
non-Roman characters, a cataloger has to choose the right character from a list of
characters -- this can be very time-consuming. With the implementation of vendor records,
the benefit of saving time for entering multiple fields is obvious.
Reducing duplicated efforts is the key for cooperative cataloging. Charlene Kellsey
describes an ideal scenario of cooperative cataloging, as she argues: the sharing of the
work of creating catalog records requires the creation of a catalog record for any given
book only once and then sharing the record with other libraries.6 As she points out the
benefit of cooperative effort in cataloging is “eliminating the duplication of effort and
diminishing the amount of original cataloging that any single library would have to do.”7
11
We have come a long way from the card catalog to machine readable records. We should
take advantage of modern technology and all the help we can get to make that ideal a
reality. Vendor records are one of the helpful methods we can take advantage of.
It has always been a concern that editing vendor records to the national
standardized records is time-consuming. Perhaps this effort would be time-consuming for
the library that edits each particular entry. But if we look at this from a cooperative
perspective, the time spent on each item would be less overall. Let’s assume a cataloger
spends 30 minutes to create an original record. If an item is held by 2 libraries, the labor
was shared by 2 and the time spent is reduced by 50%, if three, by 66% and so on. Thus,
this cooperative method will be a great benefit to all. As the data showed, the rate for
original cataloging is only 16.18%, which means we only need to spend our personnel and
time on a limited number of records. The ideal scenario of creating each item only once
and sharing this information is reachable, if we work together. It is our responsibility, not
the vendor’s, to ensure the success of a cooperative effort and the high quality of
cataloging.
To conclude, I would like to make a couple of recommendations.
1. Vendor records should not be loaded into utilities such as OCLC or any uniform
catalog or databases. In his article, Beall pointed out the negative impacts of loading non-
standardized vendor records in OCLC.8 It can result in duplicated records, duplicated
searches and duplicated editing. When a vendor record in OCLC is only enhanced at a
local level, the rest of the community will not benefit from it. This is again a cooperative
cataloging issue; the technology available now allows us to edit both non-Roman and
12
Roman data in a local system, and then save it to the OCLC local file. Thus, only complete
and standardized records will be added to the OCLC.
2. We need to educate our vendors to make better records. Some vendors have a
system to allow institutional profiles. We should establish good and detailed profiles to let
vendors know what kind of information we are looking for. Although Chinese vendor
records are only used in acquisitions, we still need to provide feedback and should
communicate with vendors for problems and issues because these information will also
serve as the descriptive part of full MARC records. We can influence vendors to establish
good practice and create quality records.
3. Further assessing vendor records. More studies like this one can give us a better
understanding of the pros and cons of these alternative cataloging methods. We will also
be able to be more specific about the positive and negative impact of using vendor records.
Based on these studies, we can find ways to adjust our workflow that would best employ
vendor records. My observations of Chinese vendor records in this study only deal with
short vendor records. It will be followed with another study on the complete MARC
records from CIBTC. I am sure this type of study and our cooperative effort will benefit
our community as a whole. In today’s library profession, librarians are, more than ever,
looking for a perfect combination of real-time speed and “meticulous quality.”9 Vendor
records will make library material available quickly, and by working cooperatively, we can
obtain this desired processing speed without jeopardizing the meticulous quality of
cataloging standards.
References
13
Beall , Jeffrey, “The Impact of Vendor Records on Cataloging and Access in Academic
Libraries,” Library Collections, Acquisitions, & Technical Services 24 (2000): 229-237.
-----, “What OCLC and RLIN Don’t Want You to Know About the Future of Cataloging,”
PNLA Quarterly 66 no. 3 (Spring 2002): 9-10.
Hanson, Heidi & John Schalow, “Two Aspects of Quality in Technical Services:
Automating for Quick Availability, and Identifying Problems, Effecting Solutions” in
Library Collections, Acquisitions, & Technical Services 23, no. 4: 433-
Huribert, Terry, Linda L. Dujmic, “Factors Affecting Cataloging Time: an In-House
Survey,” Technical Services Quarterly, 22 no. 2 (2004): 1-14.
Kellsey, Charlene, “Cooperative Cataloging, Vendor Records, and European Language
Monographs,” Library Resources & Technical Services 46, no. 3: 105-110.
Shedenhelm , Laura D. and Bartley A. Burk, “Book Vendor Records in the OCLC
database: Boon or Bane?” Library Resources & Technical Services 45 no. 1:10-19.
Wilhoit, Karen H. “Vendor Provided Technical Services: Cataloging using OCLC
PromptCat at Wright State University,” Library Collections, Acquisitions, & Technical
Services 27 (2003): 165-167.
*The author wishes to acknowledge the Research and Publication Committee of the University of Illinois at
Urbana-Champaign Library, which provided support for the completion of this research.
1
Heidi Hanson & John Schalow, “Two Aspects of Quality in Technical Services: Automating for Quick
Availability, and Identifying Problems, Effecting Solutions” in Library Collections, Acquisitions, &
Technical Services 23, no. 4: 433.
2
See Jeffrey Beall, “The Impact of Vendor Records on Cataloging and Access in Academic Libraries,” in
Library Collections, Acquisitions, & Technical Services 24 (2000): 229-237; and Charlene Kellsey,
“Cooperative Cataloging, Vendor Records, and European Language Monographs,” Library Resources &
Technical Services 46, no. 3: 105-110.
3
Laura D. Shedenhelm and Bartley A. Burk, “Book Vendor Records in the OCLC Database: Boon or
Bane?” Library Resources & Techinical Services 45, no. 1: 10-19.
4
Shedenhelm and Burk, “Book Vendor Records in the OCLC Database,” p. 15.
5
Ibid.
6
Charlene Kellsey, “Cooperative Cataloging, Vendor Records, and European Language Monographs,”
Library Resources & Technical Services 46 no. 3: 105.
7
Charlene Kellsey, “Cooperative Cataloging, Vendor Records, and European Language Monographs,” p.
105.
8
Jeffrey Beall, “The Impact of Vendor Records on Cataloging and Access in Academic Libraries,” p. 229-
237.
9
Heidi Hanson & John Schalow, “Two Aspects of Quality in Technical Services: Automating for Quick
Availability, and Identifying Problems, Effecting Solutions,” p. 433.
14