Dear Harry,
Sorry for my late response to your message.
I was out of contact for these two weeks.
The ULIS metadata schema registry where reference descriptions of
15 DC elements expressed in more than 20 languages are stored is
accessible at "http://avalon.ulis.ac.jp/registry/".
This page includes two versions.
We have been using XML and RDFS as its text encoding scheme from
the first version.
Regarding the character encoding schemes:
In the first verion, referece descriptions are primarily encoded
in local character encoding schemes, for example DC-Japanese in EUC-JP,
DC-Thai in TIS, and so on.
In the second version (=current version), Unicode (UTF8) is used as
its character encoding scheme.
Regarding user interface issues:
In the first version, a browser for the multilingual descriptions is
provided using our MHTML technology. However, no search function
is provided.
In the current version, both browsing and search functions are realized using
XSL (XSLT + XPath) and Java technologies.
(Please input "title", for example, at the text input form of the
page given above. You'll see reference descriptions of the Title elements
in many languages. IE5.0+ is recommended.)
We are not using EOR for our implementation. Our primary reason is
the character encoding restriction of the underlying database software
of EOR.
For your information, I have temporarily loaded my two papers at
http://avalon.ulis.ac.jp/~sugimoto/RPs/index.html.
The titles of the papers are
- Versioning the Dublin Core across Multiple Languages and Over Time
- A Multilingual Metadata Schema Registry Based on RDF Schema
Both of these papers are co-authored with my colleagues at ULIS and
Thomas Baker at GMD.
The first one was presented at SAINT 2001 workshop in San Diego, CA
(sponsors: IEEE-CS & IPSJ), and the second one has been accepted
for presentation at DC-2001 in Tokyo.
Related issues:
- At this moment, in the ULIS registry, a set of reference descriptions
in a particular language is stored in a single file. This is primarily
for maintenance reason. Any DB management tool is not used either for
maitenance or searching at this moment. We are seeking a "good" tool
for this purpose.
- At ULIS, we use "xml:lang" attribute to indicate a language in RDF/RDFS.
- The ULIS team is working on a metadata description in multiple languages
in a single metadata record. The metadata records are encoded in XML but
not RDF at this moment. In this metadatqa description, xml:lang is
used to indicate a language of an element in a metadata record.
We will use RDF for this system in near future.
I'm not being able to follow the discussion at the DC-Registry list well
because of my two weeks absense. Further questions or comments are
more than welcome.
Best wishes,
-- Shigeo
========
From: "Wagner,Harry" <[log in to unmask]>
To: "[log in to unmask]" <[log in to unmask]>
Cc: [log in to unmask]
Subject: FW: Internationalization
Date: Wed, 15 Aug 2001 11:12:57 -0400
Dear Sugimoto San,
Please allow me to introduce myself. I am working with Stu Weibel and
Rachel Heery on the Dublincore Registry application. I am very interested
in the presentation you gave in Budapest earlier this year (unfortunately, I
was unable to attend), and in the work you have done on the multi-lingual
registry prototype (http://www.dl.ulis.ac.jp/DCML/DCML.html). One of the
important requirements for this application is that it be multi-lingual. I
would appreciate any advice or help you could provide based on your
experience in this area.
We are currently identifying the functional requirements for version 1 of
the Registry application. Whatever the version 1 requirements are, it seems
important that we enable multi-lingual functionality in a way that will
support our long-term requirements, which should include:
- multi-lingual user interface
- ability to search the registry using any supported language
- search results displayed in different languages
I have some experience with multi-lingual web applications using '.property
files', but none using an XML approach (i.e., TMX). Can you recommend one
approach over another?
We are also struggling with how to manage the data. Should we maintain one
version of the schema, and do the language translation at the presentation
layer, or maintain schemas in each supported language? If we maintain
separate schemas for each language, are there advantages to storing them in
their own database, versus all in one database?
Would you be willing to share the source for your prototype? Any help you
can provide would be much appreciated.
Best Regards,
Harry Wagner
=========================================================
SUGIMOTO, Shigeo, Ph.D.
Professor
University of Library and Information Science (ULIS)
postal address: Tsukuba, Ibaraki 305-8550, JAPAN
phone: +81-298-59-1348(dial-in), +81-298-59-1111(ULIS)
fax: +81-298-59-1093 email: [log in to unmask]
|