It

1
Its Distributed searching, Jim

• But not as you know it!
• ZIG meeting, Leuven
• July 12th, 2000
• Bert Degenhart Drenth
• ADLIB Information Systems BV
• bert_at_nl.adlibsoft.com
• http//www.nl.adlibsoft.com

2
A problem to be solved

• Multiple database architectures
• (ODBC, ADO, other)
• Use of standard components
• Use of standard protocols
• Distributed database
• Close to Web development
• Use for cross domain application library,
  museum and archives.

3
An alternative architecture

• 3 functions
• Query through URLs (get/post method)
• Return results in XML
• Use a broker or gateway to combine results

4
A prototype

• 2 databases an ODBC source (MS-Access)an ADLIB
  database
• ODBC source contains simple library records
• ADLIB database contains Dublin Core records

5
System setup
Web browser
url
HTML
ASP broker
url
XML
XML
url
translator
translator
SQL
AQL
records
TDS
ADLIBdatabase
ODBC source
6
Technology used

• 1. Express query in URL (proprietary syntax)
• 2. Query is handed to translators by a HTTP
  request using Microsoft.XMLHTTP
• 3. Translators hand query to database engine
  using ODBC (ACCESS) or native calls
• 4. Translators return XML to broker
• 5. Broker uses XMLDOM to integrate results
• 6. Broker runs on ASP written in Jscript

7
Comparison with Z39.50

• Differences
• No BER
• No library specific technology
• Current implementation does not split between
  search and retrieve
• No extended services

8
Comparison with Z39.50

• Similarities
• Asynchronous searches possible
• For real applications standardisation needs to
  take place on
• Definition of query URLs
• The three S-s Syntax, Structure and Semantics
• Syntax XML
• Structure return elements (DTD/schema)
• Semantics attribute set and profile
• The mechanism for setting up this can be
  borrowed from Z39.50

9
Decouple transport mechanism from application

• Technology changes and is driven by forces
  outside of the library / museum / archive arena
• If you cant beat them, join them
• Inertia (MARC is still there!)

10
Developments in CIMI what happened after the
Z39.50 testbed

• Testbed proved that the technology worked, but no
  practical implementations are out there!
• No support for Z39.50 in commercial products

11
Developments in CIMI what happened after the
Z39.50 testbed

• After the Z39.50 testbed the Dublin Core testbed
  was conducted.
• Proved to be successful, with a central
  repository.
• Support for DC in XML in commercial products.

12
Developments in CIMI what happened after the
Z39.50 testbed

• CIMI moved from DC to intra community data
  exchange, based on XML.
• MDA and CIMI are producing a DTD for Spectrum,
  scheduled for november 2000
• The CIMI Z39.50 profile could be used as a basis
  for distributed applications using HTTP
  technology, notably importantrecord schema (dtd
  / XML-schema)access points and semantics

13
The end for Z39.50?

• My personal opinion NO!
• Emphasis has to shift from bits on the wire to
  the
• Three S-es for a variety of information services
• Integration, communication needs to take place
  with similar fields TEI, EAD, Spectrum-XML and
  DC
• Of course existing systems need upgrade /
  maintenance