HKUST Institutional Repository


XML and global name
access control

Ki-Tat Lam

Introduction

While there are well-established standards and
implementations for conducting authority control
at the regional level, efforts to extend such control
globally are still far from mature. For instance,
the Library of Congress (LC) has established one
of the world’s largest name authority files using
the MARC21 standard. However, this
implementation was originally designed for
Anglo-American libraries. Problems begin to
appear when users worldwide adopted LC’s
records and standards for non-Latin names. For
instance, there have been complaints about the
inability to store CJK (Chinese, Japanese,
Korean) scripts in LC’s authority records because
it is very difficult to identify who is who if the
record contains only the Latin transliterations.

Currently, the MARC21 format is a widely
adopted markup standard for authority
metadata. This standard has been very efficient
for Latin-based names. However, if it is used for
non-Latin names, the following issues may
cause difficulties:

providing more than one established
heading in the same record;
determining the field to be used to store
multiple scripts with the same name;
linking multiple fields of the same name in
a record;
identifying the script that is stored in a field;
identifying the romanization scheme used
in a field; and
determining which character encoding
scheme should be used in the record.

Enormous effort will be required to add original
scripts to LC’s authority files. However, the
most fundamental issue for supporting global
authority control lies in a search for a markup
standard that fits well with the nature of
multi-lingual name data. This article presents a
solution using XML and its related technologies
to achieve global name access control.

Using multi-scripts in MARC21 name
authority record

‘‘ ’’, the pen name of the renowned Chinese
novelist and journalist ‘‘ ’’, will be used to

The author

Ki-Tat Lam is Head of Systems, Hong Kong University of
Science & Technology Library, Kowloon, Hong Kong.

Keywords

Access control, Computer languages, Foreign languages

Abstract

This paper discusses why the MARC21-based authority

format has failed in a global setting and details the use of

XML and its related technologies to achieve global name

access control.

Electronic access

The research register for this journal is available at

http://www.emeraldinsight.com/researchregisters

The current issue and full text archive of this journal is

available at

http://www.emeraldinsight.com/1065-075X.htm

88

OCLC Systems & Services

Volume 18 . Number 2 . 2002 . pp. 88±96

# MCB UP Limited . ISSN 1065-075X
DOI 10.1108/10650750210430150

http://www.emeraldinsight.com/researchregisters
http://www.emeraldinsight.com/1065-075X.htm


illustrate the multi-script issues in name

authority records. Although this example

involves only Chinese and Latin scripts, the

problems discussed are common to other

languages and scripts. The following are the

various forms and transliterations of ‘‘ ’’:
(1) ‘‘ ’’ – pen name used for his martial arts

fiction;
(2) ‘‘Jin, Yong’’ – Pinyin romanization of this

pen name;
(3) ‘‘Chin, Yung’’ – Wade-Giles romanization

of this pen name;
(4) ‘‘Kim-Dung’’ – Vietnamese romanization

of this pen name;
(5) ‘‘Kin, Yo’’ – Japanese romanization of this

pen name;
(6) ‘‘ ’’ – name used for his non-fiction

publications;
(7) ‘‘Zha, Liangyong’’ – Pinyin romanization of

this name;
(8) ‘‘Cha, Liang-yung’’ – Wade-Giles

romanization of this name; and
(9) ‘‘Cha, Louis’’ – his English name.

Numbers 1-5 are various transliterations of the

pen name ‘‘ ’’ and numbers 6-8 are for his

real name ‘‘ ’’. The following shows the

major fields of the authority record found in the

OCLC database:

0 0 1 o c a 0 0 5 6 0 2 7 0

. . . .

1 0 0 1 | a J i n , Yo n g , | d 1 9 2 4 -

4 0 0 1 | a Ch i n , Yu n g , | d 1 9 2 4 -

4 0 0 1 | a Ki m, Du n g , | d 1 9 2 4 -

4 0 0 1 | a Ki n , Y , | d 1 9 2 4 -

4 0 0 1 | a Z h a , L i a n g y o n g , | d 1 9 2 4 -

4 0 0 1 | a Ch a , L i a n g - y u n g , | d 1 9 2 4 -

4 0 0 1 | a Ch a , L o u i s , | d 1 9 2 4 -

. . . .

This record contains many weaknesses:
It contains Latin scripts only; the Chinese

scripts, i.e. and , are not

recorded.
‘‘Jin, Yong’’ is selected as the established

heading by the Library of Congress, while

other authority control agencies may prefer

to use his other names. For example, a

Hong Kong library would use the Chinese

script in its catalog.

Presently, MARC21 can only tackle part of
these problems. It is possible to encode multi-

script by adopting the UCS/Unicode

specification (Library of Congress, 2000,
January). Position 9 in the Leader would contain
the value ‘‘a’’ to indicate that the metadata is in
UTF-8. MARC21 also offers two markup

models for multi-script records, known as
model A and model B. The following shows

how the Chinese script is handled in these two

models:

Model A
0 0 1 o c a0 0 5 6 0 2 7 0

. . . .

1 0 0 1 | 6 8 8 0 - 0 1 | a J i n , Yo n g , | d 1 9 2 4 -

8 8 0 1 | 6 1 0 0 - 0 1 | a , | d 1 9 2 4 -

4 0 0 1 | a Ch i n , Yu n g , | d 1 9 2 4 -

4 0 0 1 | a Ki m, Du n g , | d 1 9 2 4 -

4 0 0 1 | a Ki n , Y , | d 1 9 2 4 -

4 0 0 1 | 6 8 8 0 - 0 2 | a Z h a , L i a n g y o n g , | d 1 9 2 4 -

8 8 0 1 | 6 4 0 0 - 0 2 | a , | d 1 9 2 4 -

4 0 0 1 | a Ch a , L i a n g - y u n g , | d 1 9 2 4 -

4 0 0 1 | a Ch a , L o u i s , | d 1 9 2 4 -

. . . .

Model B
0 0 1 o c a 0 0 5 6 0 2 7 0

. . . .

1 0 0 1 | a J i n , Yo n g , | d 1 9 2 4 -

4 0 0 1 | a Ch i n , Yu n g , | d 1 9 2 4 -

4 0 0 1 | a Ki m, Du n g , | d 1 9 2 4 -

4 0 0 1 | a Ki n , Y , | d 1 9 2 4 -

4 0 0 1 | a Z h a , L i a n g y o n g , | d 1 9 2 4 -

4 0 0 1 | a , | d 1 9 2 4 -

4 0 0 1 | a Ch a , L i a n g - y u n g , | d 1 9 2 4 -

4 0 0 1 | a Ch a , L o u i s , | d 1 9 2 4 -

7 0 0 1 | a , | d 1 9 2 4 -

. . . .

By adopting tag 880 and subfield 6 in model A,

it is possible to:
make parallel to the established form
‘‘Jin, Yong’’; and
maintain a link between and its
Pinyin form ‘‘Zha, Liangyong’’.

MARC21 specifies that the transliteration is
stored in the regular tag with its vernacular

forms stored in multiple 880 tags. With a slight
modification to MARC21 by exchanging the

89

XML and global name access control

Ki-Tat Lam

OCLC Systems & Services

Volume 18 . Number 2 . 2002 . 88±96



regular tag and tag 880, model A will be able to
link all transliterations of the same name form:

1 0 0 1 | 6 8 8 0 - 0 1 | a , | d 1 9 2 4 -

8 8 0 1 | 6 1 0 0 - 0 1 | a J i n , Yo n g , | d 1 9 2 4 -

8 8 0 1 | 6 1 0 0 - 0 1 | a Ch i n , Yu n g , | d 1 9 2 4 -

8 8 0 1 | 6 1 0 0 - 0 1 | a Ki m, Du n g , | d 1 9 2 4 -

8 8 0 1 | 6 1 0 0 - 0 1 | a Ki n , Y , | d 1 9 2 4 -

In model B, tag 700 (established linking entry)
is used to store the Chinese script , making
it an alternate established form. However,
model B fails to maintain the parallel
relationship between multiple transliterations of
the same name. The two fields for and
‘‘Zha, Liangyong’’ are not linked. Therefore,
one can fall back from model A to model B, but
not from model B to model A. Although model
A allows better linking of the vernacular script
with its transliterations, neither library
automation vendors nor bibliographic utilities
such as OCLC are eager to support it for
authority control. None of these models are
sufficient for use in the global setting.

Three areas in the MARC21 authority format
could benefit from enhancement. Firstly, 1XX
should be repeatable, so that instead of
depending on 7XX, more than one name can
be coded in 1XX as established forms.
Secondly, better linking capability between
related forms of the name should be developed
to avoid using tag 880 with subfield 6. This is
because library automation vendors are
reluctant to implement tag 880 for authority
control, a task that requires substantial effort.
Thirdly, there should be a place to hold the
multi-lingual attributes of each name, including
information about the script, language and
romanization scheme. Although effort is needed
to identify these attributes, and it is in some
cases not an easy task, such information is
particularly helpful in differentiating names in a
multi-lingual environment.

From MARC to XML

XML (eXtensible Markup Language) is a
simplified standard of SGML. It allows plain
text data to carry not just layout information,
but also semantic structure. Since its official
release in early 1998 by the World Wide Web

Consortium (W3C), XML has rapidly been
adopted and implemented by many industries.
For the past few years, many initiatives and

projects that make use of XML for library
metadata have emerged (e.g. bibliographic,
authority, holdings, patron records).

MARC has been widely used as a markup
language for library metadata since the1960s.
MARC21, UNIMARC and their variations can
be considered as MARC’s DTD (Document

Type Definition) or markup schema. Due to
the similarity between markup concepts in
XML and MARC, it is feasible to use XML to

describe a MARC record. By converting
MARC to XML, library applications would be
able to exchange library metadata in the
emerging XML format and avoid working with

the proprietary MARC interchange format,
known as ISO2709.

In addition, using XML, XSLT (Extensible

Stylesheet Language Transformation) and
other Web technologies, it is possible to do the
following:

create library metadata once and publish it

in different formats;
view library metadata directly from Web
browsers, search engines, and potentially,

library automation systems, without
conversion;
convert robustly between XML and MARC
formats without data loss; and
resolve many problems that were inherent
with the MARC format.

In 1995, the Library of Congress began to look
into the feasibility of using SGML to encode

USMARC (a previous version of MARC21)
data (Library of Congress, 2001) in the
subsequent release of MARC DTDs and
conversion software. In 1999, Lane Medical

Library at Stanford University released its
XMLMARC software, a byproduct of its
Medlane Project, for public use (Lane Medical

Library, 2002). Based on its in-house
developed DTDs for bibliographic and
authority formats, this java-based XMLMARC
program offers conversion of MARC21 to

XML. Many library automation vendors also
developed their own DTDs for their MARC
data. For example, Innovative Interfaces Inc.

90

XML and global name access control

Ki-Tat Lam

OCLC Systems & Services

Volume 18 . Number 2 . 2002 . 88±96



has its own DTDs to handle various record

types.
All these projects are defining their own XML

schema (DTDs). However, to facilitate wider

acceptance, it is essential that a globally agreed

upon XML schema for MARC metadata be

developed. This global MARC-XML standard

should closely follow the MARC structure. In

addition, it should allow for variations, changes

and enhancements in the content designation.
Unlike the DTDs developed by LC and the

Lane project, which are tightly based on

MARC21, the global MARC-XML elements

and attributes should be independent of

regional content designation standards. To

achieve this, Figure 1 shows the proposed

namespace.
In the example, <ma r c > and </ ma r c > mark

the beginning and ending of a MARC record.

<f d > and </ f d > mark the beginning and ending

of a MARC tag. <s f > and </ s f > mark the

beginning and ending of a subfield. Elements

f d and s f contain a number of attributes. The

name attribute defines the MARC tag name or

the subfield code. i n d 1 and i n d 2 are the two

indicators of the MARC tag. There should be

additional attributes for f d and s f , to encode

information not found in MARC. For instance,

we can have:
i d – to give a unique id to each MARC tag

and to allow for linking among MARC tags;
s c r i p t – to encode the language, the script

and the romanization scheme used in the

MARC tag;
l a b e l – to describe the MARC tag, etc.

Attributes for the ma r c element can also be

defined to encode record level information

found in the leader and the controlled fields

(e.g. the 00X tags in MARC21).

Global name access control in XML – an
experiment

With the adoption of XML, it is possible to
expand authority control to access control. The
idea of access control, as opposed to authority
control, is to make obsolete the notion of
‘‘authority’’ so that variant forms of a name are
equal in status and users are able to select any of
them for searching, retrieval and display.

Based on the proposed MARC-XML schema
and the MARC21 specification, the name
access control metadata for ‘‘ ’’ in XML is
shown in Figure 2.

Note that all transliterations for the name
‘‘ ’’ are assigned a value of ‘‘100’’ for their
name attribute. And for linking, the i d
attributes of these f d elements have the same
number stem (e.g. 1.1, 1.2, ). Similar
assignments are applied to the name
‘‘ ’’. Also note that the multi-lingual
information is stored in the s c r i p t attribute,
in the form of:

" s c r i p t . l a n g u a g e . r o ma n i z a t i o n "

A feasibility study on the above MARC-XML
format was conducted at the Hong Kong
University of Science and Technology
(HKUST) Library in the summer of 2001. A
prototype system for a Global Name Access
Control Metadata Repository was developed.
The system was built on the XML-based
Tamino Server, using the Perl programming
language and XSLT stylesheet transformation
to generate search results and displays. Figures
3-6 show MARC21 extended pages of the full
record, and sample screens of the system: the
main page, the search results page, and the
public display.

Note that in Figure 3, the prototype system is
capable of converting the MARC-XML

Figure 1 Proposed namespace

91

XML and global name access control

Ki-Tat Lam

OCLC Systems & Services

Volume 18 . Number 2 . 2002 . 88±96



metadata to various MARC21 formats,

including model A and model B. It should also

be possible to convert the MARC-XML format

to the UNIMARC format.
It is necessary to extend MARC21 so that the

globalization attributes contained in the

MARC-XML format can be maintained after

the mapping. This can be achieved by

enhancing subfield 8 (Field Link Control

Subfield) for storing the i d and the s c r i p t

attributes, for example:

| 8 1 . 2 \ s \ l a t i n . Ch i n e s e . p i n y i n

‘‘s’’ (script) is a new field link type that

differentiates the multi-lingual attributes of the

linked fields. See Figure 6 for the use of

subfield 8.

MARC-XML offers other added-value
features. It will be possible to implement

Barnhart’s (1996) idea of reducing data
content redundancy in a name access control
record by separating and linking data elements
such as uniform title and dates associated with
the name from the name elements.

With sufficient globalization information
encoded in the MARC-XML name metadata, a
virtual international authority file as proposed
by Tillett (2001) can become a reality. It is

feasible to have globally distributed repositories,
offering name access control metadata as Web
services, for access by various client

applications such as OPACs (Online Public
Access Catalogs), authority control modules,
bookstore catalogs and rights management
systems.

Figure 2 An example of name access contol metadata for ÁÁ ÂÂ in XML

92

XML and global name access control

Ki-Tat Lam

OCLC Systems & Services

Volume 18 . Number 2 . 2002 . 88±96



SOAP communication with the repository

SOAP (Simple Object Access Protocol) (W3C,
2001) is an emerging standard of the W3C. It is
a simple protocol for information exchange in a
distributed environment. Using this standard,
the prototype name access control metadata

repository can be published as a Web service. A
Web service is a collection of functions that are
published on the Web for use by other
programs (Glass, 2000).

To verify the SOAP communication concept,
the prototype repository was enhanced as a
SOAP node. Another SOAP application,
pretending to be an INNOPAC WebPAC
interface that communicates with the repository
for author search assistance, was also
developed. Figure 7 shows the SOAP
messaging flow of this setup.

We chose to use the SOAP technology
instead of the Z39.50 standard in this project.
Although Z39.50 is a well-established
information retrieval standard for library
applications, it was designed before the
World Wide Web era. Because of the steep
learning curve for Z39.50, developers need to
invest extra effort in order to deploy and
support Z39.50 systems. On the contrary,
SOAP uses the two common standards XML
and HTTP for messaging and transport.
Because it introduces no new technology and
is simple to follow, SOAP can quickly be
deployed for distributed applications
requiring remote procedure calls. Driven by
the advantages of these Web technologies, a
group of Z39.50 implementers has begun to
study the integration of XML, HTTP and
SOAP to the Z39.50 standard (ZNG
Initiative, 2001).

Figure 8 shows some SOAP messages
between the INNOPAC and the repository.

It is desirable that a globally agreed-upon
profile for these kinds of SOAP request and
response messages be formed. Once the
specification is in place, distributed authority
repositories can be published as compatible
Web services, serving MARC metadata in XML
format. Similar to the idea of Z39.50, client
applications can then communicate with
selective repositories worldwide for their name
access control.

Conclusion

There are limitations and problems in
MARC21 for name access control at the global
level. These problems can be resolved with the

Figure 3 Main page of the metadata repository system

Figure 4 Search results page of the metadata repository system

93

XML and global name access control

Ki-Tat Lam

OCLC Systems & Services

Volume 18 . Number 2 . 2002 . 88±96



help of XML and related technologies. The
proposed MARC-XML schema can crosswalk
with MARC21 and any other MARC content
designation versions, such as UNIMARC.
With the use of SOAP, XML-based name
access control metadata can be easily

published as Web services. If major authority

files, such as the Library of Congress name

authority file, are converted to the proposed

XML format and are enabled as SOAP nodes,

a true global name access control environment

will be achieved.
To make the above scenario happen, a

global effort should be established to specify

two standards, namely the XML schema for

the MARC format and the SOAP messaging

profile for bibliographic information

retrieval.

Figure 5 OPAC format view of the metadata repository system

Figure 6 MARC21 extended format view of the metadata repository system Figure 7 SOAP request and response

94

XML and global name access control

Ki-Tat Lam

OCLC Systems & Services

Volume 18 . Number 2 . 2002 . 88±96



References

Barnhart, L. (1996), ÁÁAccess control records: prospects and

challengesÂÂ, Authority Control in the 21st Century: An

Invitational Conference, March 31-April 1, 1996,

available at: www.oclc.org/oclc/man/authconf/

barnhart.htm (accessed 29 January, 2002).
Glass, G. (2000), ÁÁThe Web services (r)evolution, part 1:

applying Web services to applicationsÂÂ, available at:

www-106.ibm.com/developerworks/webservices/

library/ws-peer1.html (accessed 29 January, 2002).
Lane Medical Library, Stanford University Medical Center

(2002), ÁÁMedlane XMLMARCÂÂ, last updated

7 January, available at: xmlmarc.stanford.edu/

(accessed: 29 January, 2002).
Library of Congress (2000), ÁÁMARC 21 specifications for

record structure, character sets, and exchange media.

Character sets: part 2. UCS/Unicode environmentÂÂ,

January, available: http://lcweb.loc.gov/marc/

specifications/speccharucs.html (accessed:

29 January, 2002).
Library of Congress (2001), ÁÁMARC SGML and XMLÂÂ, last

updated March, available at: lcweb.loc.gov/marc/

marcsgml.html (accessed 29 January, 2002).
Tillett, B.B. (2001), ÁÁA virtual international authority fileÂÂ,

67th IFLA Council and General Conference, August

16-25, 2001, available at: www.ifla.org/IV/ifla67/

papers/094-152ae.pdf (accessed 29 January, 2002).
W3C (2001), SOAP Version 1.2 working drafts, available

through the links in: http://www.w3.org/TR/2001/WD-

soap12-part0-20011217/ , December 17 (accessed

29 January, 2002).
ZNG Initiative (2001), ÁÁZ39.50 next generationÂÂ, July,

available at: www.loc.gov/z3950/agency/zng.html

(accessed 29 January, 2002).

Implications for practitioners

This summary has been provided to allow a
rapid appreciation of the significance of
the content of this article. Browsers may
then choose to read the article in toto, to
derive full benefit from the authors’ work.

It is often said that communications and
technology have combined to make the world a
smaller place.

Such ideas, and with them the concept of the
global village, are attractive. However, they are not
much consolation to library professionals faced with
issues of global name access control.

The Library of Congress used the Machine
Readable Cataloging (MARC) 21 standard to
establish one of the world’s largest name authority
files. It was designed for Anglo-American libraries
but its use has been far wider.

So, it is when some worldwide users adopt the
Library of Congress records and standards for
non-Latin names that problems occur. Chinese,
Japanese and Korean are just three examples of
scripts which create difficulties for MARC21 in
Congress authority records, because of the
limitations imposed by Latin transliterations.

Yong Jin, the Chinese novelist and journalist,
serves as a good example. That appellation
would be the entry taking the Pinyin romanization of
his pen name for martial arts fiction. One could
make eight more entries, including the Chinese script
version of that name to Japanese and Vietnamese
romanizations to his English name.

There are many weaknesses in the major fields of
the authority record found in the OCLC database,
and MARC21 can at the moment handle only some
of them. The writer provides two different models
which find ways of addressing some of the issues.
Among them is the Established Linking Entry
created by one model. This stores the Chinese script
and makes it an alternate established form. The
other allows better linking of the vernacular script
with its transliterations.

However, bibliographic utilities such as OCLC,
to take just one example, would not want to support
it for authority control. Neither model is sufficient
for use in a global setting; something more radical is
needed.

Extensible Markup Language (XML) and
related technologies can help. XML has been

Figure 8 SOAP messages between the INNOPAC and the repository

95

XML and global name access control

Ki-Tat Lam

OCLC Systems & Services

Volume 18 . Number 2 . 2002 . 88±96

http://www.ifla.org/IV/ifla67/papers/094-152ae.pdf
http://lcweb.loc.gov/marc/specifications/speccharucs.html
http://www.oclc.org/oclc/man/authconf/barnhart.htm
http://www.oclc.org/oclc/man/authconf/barnhart.htm
http://www-106.ibm.com/developerworks/webservices/library/ws-peer1.html
http://www-106.ibm.com/developerworks/webservices/library/ws-peer1.html
http://lcweb.loc.gov/marc/specifications/speccharucs.html
http://www.ifla.org/IV/ifla67/papers/094-152ae.pdf
http://www.w3.org/TR/2001/WDsoap12-part0-20011217
http://www.w3.org/TR/2001/WDsoap12-part0-20011217
http://www.loc.gov/z3950/agency/zng.html


rapidly adopted and implemented by many
industries during the last four years. XML and
other Web technologies offer some answers,
including conversion between XML and MARC
formats without data loss and with resolution of
many problems inherent in MARC.

What we need now is a globally agreed-upon use
of these opportunities. A MARC-XML feasibility
was conducted last year at the Hong Kong
University of Science and Technology Library. It
suggests that, through globalization information
encoded in the MARC-XML name metadata, a
virtual international authority file can become a
reality.

Other areas for further investigation include
Simple Object Access Protocol (SOAP), an emerging

World Wide Web Consortium standard. SOAP

carries the dual advantage of being easy to follow

while introducing no new technology. It could serve

MARC metadata in XML format, and its use could
pave the way for XML-based name access control

metadata to be published as Web services.
The ultimate aim has to be creation of a true

global name access control environment. That

requires a global effort to work towards the
standards described here.

That shrinking world or global village, which

nevertheless is witness to an information explosion,

demands nothing less.

(Précis supplied to MCB UP Limited by

consultants.)

96

XML and global name access control

Ki-Tat Lam

OCLC Systems & Services

Volume 18 . Number 2 . 2002 . 88±96