Semantic Web:

1
Semantic Web To be browsed or what?

• Martin Dzbor
• Knowledge Media Institute, The Open University

Presentation _at_ Soc. of Archivists Conf. 2nd Sep
2004
2
Overview

• What is Magpie?
• Brief demonstration
• Principles assumptions
• What can it do and how?
• Supporting document interpretation
• Collaborative browsing
• Conceptual management of browsing history
• Application areas
• Education e-learning
• Access knowledge not content

3
Semantic Services Demo(Magpie in education)
4
Semantically Enhanced Web Browsing

• Zero overhead costs
• Dynamic ontology-driven enrichment in real time
• Utilizing interfaces familiar to the lay user
• Resource re-usability openness
• Gateway to the semantic web
• Access knowledge through annotated web resources
• Integrated with standard web browsers
• Magpie services are first class concepts
• Multiple user interaction modalities
• User-requested (pull or clickgo) semantic
  services
• Trigger (push or subscribelearn) semantic
  services
• One-click (or less) interface

5
Magpie Components
6
Value of tools like Magpie

• Quan Karger (2004)
• separate pieces of information about a single
  resource that used to require navigation through
  different web sites can be merged together
  without specialized portal sites or coordination
  between the sites
• McDowell, Etzioni al. (2003)
• How do we entice non-technical people to
  structure their data?
• semantic annotation will be motivated by
  services giving immediate benefits instant
  gratification to the user.
• Tauscher Greenberg (2001)
• lt 1 browsing actions use explicit history
  mechanisms
• Browsing history needs better representation that
  would be meaningful to the user.

7
and now I put my learning hat on
8
Benefits to a student

• Science (but also art philosophy) is largely
  about relating things and seeing patterns
• Quick access to course-related resources
• Ability to link theoretical resources with the
  practical articles, documents, stories,
• Exploratory guiding as a metaphor
• What would my prof say if he was watching over
  my shoulder when I read about IPCC climate
  scenarios?
• Uni may re-use chunks of existing knowledge from
  third parties and only provide their
  interpretation rather than replicating them

9
Collaborative Work Demo
10
What else?

• Conceptual management of browsing history and
  semantic bookmarks

11
Benefits of semantic bookmarking

• Textual vs. iconic representation of browsing
  history
• Automated creation of a semantic footprint in the
  selected ontology for each visited page (if the
  user allows)
• High-level semantic filtering/querying
• Yet-another gateway to the semantically
  relevant knowledge/information (via Magpie
  on-demand services)
• Comes for free with the Magpie framework

12
Current future work

• Better support for semantic web services
• Modularization of semantic services
• Tools supporting non-IT authors in publishing new
  services for their users/customers
• Service discovery composition
• Early-stage Authors toolkit

13
(No Transcript)
14
Current future work

• Better support for semantic web services
• Magpie as a development framework for semantic
  web applications
• S199 Climatology course pilot for The Open
  University its spin-off for climateprediction.ne
  t project with gt60k users
• ELS Semantically augmented environment for
  journal subscribers (for Nature Publishing)
• WIND Simple semantic web demonstrator for
  Italian Seria A football domain (variants for
  rugby, politics, )

15
Summary

• Semantic web browsing to support sense-making
• Integrated with standard web browsers
• User requested (pull) and trigger (push) semantic
  services
• Semantic web browsing at no-extra-cost and
  no-extra-cognitive-overhead for the user
• Suite of services complementing the core engine
• Team/group collaboration
• Browsing history management
• Rich educational resource appl. development
  framework
• Not a semantic web browser but browser or
  better gateway to the semantic web

16
Thank you!

• Magpie IE Plugin downloadable from
• http//kmi.open.ac.uk/projects/magpie

17
(No Transcript)
18
Elements of successful collaboration

• Common goal to provide focus/motivation
• Shared processes and workflow
• Integration of groupware into main info flows
• Critical mass
• Active presence, low use insufficient data
• Burden/benefit ratio
• Users buy-in, effort overhead costs
• Trust
• Reluctance to cross-organizational sharing
• Team rewards and recognition
• Individual performance vs. teamwork conflict?

After Tamara Hall, MITRE Corp. Intelligence
Community Collaboration (1999)
19
Magpie as a collaborative framework

• Collaboration integrated into daily activities
  that are normally centred on perusing web
  documents
• Zero-cost paradigm of the framework significantly
  reduces the burden/benefit ratio
• Automatically populated ontology (via IE screen
  scraping) overcomes bootstrapping and data
  seeding issues (a.k.a vicious circle of
  critical mass)
• Open architecture facilitates integration with
  existing IT infrastructure (and hence existing
  work practices)
• Reconciliation of individual vs. team conflict
  e.g. by requesting specific info or tapping into
  shared Memex-style repository

Chakrabarti, S., Srivastava, S. al. (WWW2000)
20
Functional Collaboration Components

• Semantic annotations automatically updated
  semantic log team memory
• Team members benefit from colleagues semantic
  annotations and browsing histories
• Synchronous communication active presence using
  a KMi lightweight telepresence/CSCW tool
  (BuddySpace)

21
ExploitIT.com
22
Current and Future Work

• Pending applications
• Climate Modelling www.climateprediction.net
• WIND - Italian Seria A
• Nature - Life Sciences Encyclopedia
• Integrating NER
• Integrating semantic web services
• Integrating semantic annotation
• Creating a framework for web site development
• Smarter lexicon generation

23
Recent, current future work

• Beyond ontology driven named entity recognition
• Remove brittleness of ontologies
• Facilitate knowledge acquisition
• Pre-alpha prototype combining integrating
  Magpie eSpotter functionalities

24
ESpotter Demo
25
Recent, current future work

• Beyond ontology driven named entity recognition
• Tighter integration with semantic web services
• Further modularization of semantic services
• Support for publishing, discovery composition
  (?)
• Automated management of Magpie services for
  different ontologies, users browsing patterns
• Early-stage conceptual wrapper for IRS-II

26
ClimatePrediction.net
Double CO2 production
Control
1994
1994
27
ClimatePrediction.net
Double CO2 production
Control
2094
2094
28
Magpie IE Plugin
Web Browser Interface (incl. click management)
Trigger Services Interface (GUI-s)
u
a
c
Ontology visualizer/browser
Text highlighting (incl. on-demand services)
l
t
d
User preferences about actions
Recognized entities (hash)
r
e
l
1
Selected classes/branches
Entity recognition interface
q
g
s
1
Lexicon-based
Rule-based (eSpotter)
Ontology browsing, mapping module
f
h
h
j
i
Document Object Model
p
o
Ontology cache
Rules patterns
o
k
Hierarchical rep. suitable for simple reasoning
k
n
Edit, manage, extend/learn or customize the rules
RDF(S)/OWL parsers
m
Remote ontologies, KB-s services