The KB on its way to Web 2.0

1
The KB on its way to Web 2.0

• Lower the barrier for users to remix the output
  of services.
• Theo van Veen, ELAG 2006, April 26

2
Overview

• Goals
• Trends in information retrieval
• How do we enhance integration
• Explanation of the concept of service integration
• Demonstration of one solution (Ajax)
• What data and service providers can do
• Remaining issues

3
Goals

• Improve sharing of information and services
• Bring enriched information within reach of a
  mouse click
• Lower implementation barriers to create new
  functionality by combining existing services
• Enhance personalization user chooses which
  services to integrate

4
Some trends in information retrieval

• Web 2.0 Let users (re)mix information and
  services from others and themselves in their own
  way
• Ajax (Asynchronous, Javascript and XML)
  Simultaneous requests can be sent from a web page
  to multiple targets. Results are retrieved
  asynchronously without freezing the screen and
  are dynamically transformed (and mixed) to new
  web pages.

5
How to enhance integration

• By
• standardisation of access to services
• publishing standard service descriptions
• creating services registries
• But also by
• describing the non-standard world as it is
• to make use of what is available now
• to facilitate the standardisation process
• combining (the best of) different native
  standards to evolve into new ones
• describing relations between services and the
  metadata that might trigger those services !

6
Extrapolation of OpenURL concept

• OpenURL
• Vendor specific knowledge base how to access
  services
• User is linked to OpenURL resolver providing
  links
• Links to services are controlled by users
  institution
• Towards (expected)
• Service providers publish all their services
• User selects preferred services stored in a
  personal knowledge base
• User controlled criteria for invoking the service
• Direct linking to service, automatic or on user
  request, based on users preferences, skipping
  the OpenURL resolver

7
The components of a new concept

• Any web page, for example showing results of a
  search
• The services any application that can be invoked
  by an URL
• The user agent, which can be
• A browser extension (Firefox)
• A portal running in the browser (Ajax)
• A server side application that intercepts user
  requests and service responses
• An OpenURL link resolver
• A knowledge base containing information on
  services, how to access those services, which
  metadata should trigger a link and the user
  preferences with respect to handling services

8
Schematic illustration of the concept
Output from service A
Output from service A gets link to service B
Service B
Description of service B
4. Modify presentation and add links e.g. to
service B
Service A
9
Demonstration of concept add image search for
creator field
lth3gtHTMLltp/gt This html pages demonstrates how
the field "ltcreatorgtShakespearelt/creatorgt" can be
processed by a user agent. lt/h3gtltbr /gt
10
COinS (Context Object in Spans)

• HTML tag ltspangt to be recognized as OpenURL
  metadata object. Example

ltspan class"Z3988" title"ctx_verZ39.88-2004amp
rft_val_fmtinfoofi/fmtkevmtxjournalamprft.
issn1045-4438"gtlt/spangt

• Usage mainly OpenURL resolution in combination
  with browser extension (bookmarklet,
  greasemonkey) that can change the behavior of the
  ltspangt
• Benefits works in most browsers because it makes
  use of standard HTML tags and attributes
• Disadvantage assumes OpenURL aware services

11
Normal response form Google Scholar
12
OpenURL link added by user agent to Google
Scholar response
13
(semi-)automatic invocation of a SRU search
14
When found the link is changed
15
Activation of another link
16
When it is NOT found the link remains an OpenURL
link
17
Ajax Asynchronous, Javascript and XML
Native protocol
gateway
SRU
SRU
XSL and Javascript
Use XSL and Javascript to create a search page
Transform XML collection descriptions using XSL
User query is sent via SRU to all targets ...
Z39.50 targets will use a Z39.50-SRU gateway
For Z39.50 not under our control ...
a central gateway will be provided
Add your own functionality and collection
descriptions
Integration is in your browser!
TEL
XML collection descriptions (SRU)
18
The knowledge database

• Combination of
• Machine readable descriptions of service behavior
  and how to access services
• URL
• URL syntax
• Request parameters
• Fixed parameters
• User interaction
• Access mechanism (POST, GET, SOAP)
• Services in relation to metadata elements
• Which metadata field triggers which services?
• How does it trigger the service?

Demo of Ajax portal
19
(No Transcript)
20
(No Transcript)
21
(No Transcript)
22
(No Transcript)
23
(No Transcript)
24
(No Transcript)
25
(No Transcript)
26
With a little bit of imagination (work to be done)

• Services can be triggered by more complex
  criteria rather than the presence of a single
  metadata field in a structured metadata record
• Services can be activated in the background and
  only appear when there is something to show
  rather than bothering the user in advance
• Services can trigger other services, for example
  from location to coordinates to display on a map
• Web pages can be analyzed (semi-automatically) to
  discover potential services and generate new
  service descriptions to be added to the service
  descriptions (like favorites)
• By formalization of service descriptions users or
  user agents may exchange working service
  descriptions

27
From service description to development of a
standard
Describing non-standardized services in a
knowledge base will facilitate the
standardization process and the integration
(automatic invocation) of services.

• Describe input and output of different services
  that do the same thing
• Describe common parameters
• Recognize parameters that describe the
  differences between services and their extra
  options
• A good generic formal description of different
  services with native protocols almost specifies a
  standard or a universal interface to these
  services

28
What can data/service providers do?

• Provide machine readable output (XML)
• Semantic tagging of HTML pages to recognize
  metadata in web pages (e.g. COinS, unAPI)
• Prevent the need for complex interpretation of
  output, be tolerant with respect to input to make
  it easy to use output from services and to
  generate input into services
• Use existing standards or create new ones
  analogue to existing ones (e.g. SRU)
• Provide service descriptions for any service that
  is usable in this context at a standard location
  e.g. http//your.host/services.xml and
  searchable by Google
• Create registries with generic service
  descriptions

29
Issues

• Are providers willing to provide machine readable
  data without branding?
• Even with semantic tagging there is a possibility
  that user agents will hide branding information
• Still much variation in the encoding of metadata
• Security issue with mixing and merging services
  from different domains in the browser
• Providers may not be aware that they are offering
  services that are useful for integration
• Providers and institutions may not yet be ready
  for Web 2.0 (protection of their data)

30
Thanks Email theo.vanveen_at_kb.nl