Low-cost semantics-enhanced web browsing with Magpie

1
Low-cost semantics-enhanced web browsing with
Magpie

• Enrico Motta
• Knowledge Media Institute
• The Open University, UK

2
Structure of the talk

• The two minutes introduction to the semantic web
• HLT for the Semantic Web Low cost, high value
  solutions
• A little semantics goes a long way (Jim
  Hendler)
• A little semantics a little HLT goes even
  further (Enrico Motta)
• Semantic web browsing with Magpie
• Conclusions

3
What is the semantic web?
4
(No Transcript)
5
(No Transcript)
6
(No Transcript)
7
Advantages of semantics (1)

• Keyword Search -gt semantic search
• Promise of increased precision

8
Demo Semantic Search vs Keyword Search
9
Ontologies HLT Understanding
Academic hasAffiliation hasAddress
hasWebAddress...
Peter Scotts Home Page
Linguistic vs Conceptual Interoperablity
10
Aqualog
11
Agents on the Web
Please get me an appointmentwith a dealer nearby
which sells Naim Royd for Saturday morning
Enricos Semantic Agent
12
Basic components of a semantic web

• Infrastructure
• Languages
• Ontologies Repositories
• Ontology Learning
• Annotation tools
• Automatic extraction of semantic content
• MnM, Melita, Armadillo, OntoMat, Annie, ESpotter
• Semantic Web Services Platforms
• Basic Functionalities
• Semantic Search
• Question Answering
• Brokering and composition of semantic web services

13
Applying HLT to Semantic Web What works and
what does not
14
(No Transcript)
15
The trouble with MnM

• Manual Annotation is a slow process
• Not clear what benefits one gets from annotation
• Chickenegg problem
• Automated annotation in MnM requires a learning
  phase and manual markup
• No free lunch for users
• Training phase requires HLT skills
• Overall cost/benefit tradeoff is negative

16
Magpie Getting an excellent cost/benefit ratio
17
Low Cost Semantics

• Focus on ontology based semantic enrichment of
  web pages
• Browsing -gt Semantic Browsing
• Zero cost for end user
• Zero work (for end user)
• Zero time overhead
• Zero cognitive overhead
• Integrated with standard web browsers
• Minimal changes to look and feel of web pages
• One-click (or less) interface

18
More about Magpie

• User requested (pull) and trigger (push) semantic
  services
• Simple but effective use of HLT technology

19
(No Transcript)
20
(No Transcript)
21
(No Transcript)
22
(No Transcript)
23
(No Transcript)
24

• A iconic representation of browsing history
• B syntactic neighbourhood of a web page
• C semantic footprint of a web page in selected
  ontology
• D high-level semantic filtering/querying of
  browsing history

25
Architecture (old)
26
Application areas

• Semantics-assisted browsing
• Pick an ontology and off you go!
• Educational tool
• Ontology-based interpretation of scientific
  publications and resources
• Web browsing independent from the document
  authors intention customizable by uni/college
• Many applications being discussed
• Science, semweb community, ISP providers,
  publishers, etc..

27
(No Transcript)
28
(No Transcript)
29
Supporting Collaborative Work

• Semantic annotations semantic log serve as a
  team memory
• Team members benefit from colleagues semantic
  annotations and their browsing history
• Synchronous alerts and communication using a KMi
  Jabber/XML infrastructure (BuddySpace)

30
Things to do (1)

• Integration with semantic web services
• IRS-II platform
• Turn the key solution
• Open to ontology languages and services
• One click select ontology and generate lexicon
  facility
• Integration with generic named entity recognition
• Integration with mini-ontology browser
• Ability to use existing semantic annotations
• Integration with annotation tools

31
Things to do (2)

• Magpie as web site specification framework
• Versions for other platforms
• Currently only IE on PCs
• Magpie-assisted search
• Magpie-Google - launch Google and then filter
  results through magpie
• Magpie-crawling - crawler finds pages related to
  current ontology Magpie robot that finds web
  pages which make sense for ontology
• Collaboration Customization

32
Web Resources
Ontologies
Semantic Services
Lexicon
33
Conclusions

• Focus on usability and integration of semantic
  and language technologies
• Low cost solutions
• Semantic web provides a great new playground for
  HLT research
• However old problems get new twists
• Magpie is an example of a low-cost, low tech
  integration of HLT and semweb technology, which
  appears to work

34
(No Transcript)