The Semantic Web and Research Challenges

1
The Semantic Web and Research Challenges

• by Tim Berners-Lee
• MIT Laboratory For Computer Science
• Director, World Wide Web Consortium
• January 2003

2
Overview

• 1. What is it/How to explain it.
• Slides 3-26.
• 2. Some things we have been doing at MIT/LCS.
• Slides 27-35.
• 3. Challenges for Research.
• Slides 36-44.
• 4. More information.
• Slide 46.
• 5. Slides not used.
• Slides 47-55.

3
1. What is it/How to explain it

• Machine processable data.
• Extending relationship data to a web.
• Solving the Application Integration problem.
• The infrastructure for the next IT revolution.

4
Everything has a URI

• Don't say "color" say "http//www.pantomime.com/20
  02/std6color"
• In N3,
• _at_prefix pan lthttp//www.pantomime.com/2002/std6gt
  .myCar pancolor "blue246".

5
The relational database
6
The element of the Semantic Web

• Can be encoded in XML.
• Simplicity and mathematical consistency.
• This is called Resource Description Framework
  (RDF).
• In N3,
• _at_prefix pan lthttp//www.pantomime.com/2002/std6gt
  ._at_prefix my ltgt.mycar pancolor "456".

7
Semantic web includes tables,...
8
...trees
9
... everything
10
RDF data..
11
...merges just like that.
12
RDF Semantic links - "Joining the Web"
13
Enterprise Application Integration problem
14
RDF Application Integration hub
15
3. More expressive power
16
RDF data layer -goals and status

• Goals
• Data serialization format.
• Inter-application interoperability across
  applications.
• Status
• RDF is W3C recommendation. Second cycle in
  progress.

17
RDF Schema layer - technology

• Minimalist model - (thing), Class, Property.
• Subproperty, Subclass.
• Domain Range.
• Comments labels.
• Very wide interoperability.
• All that is needed for interoperability of the
  vast amount of "data" on the web.

18
RDF Schema layer - goals and status

• Goals
• Allows vocabulary definition.
• Can go ahead and make schemas for existing and
  new applications.
• Status
• RDF Schema is W3C Last Call WD, very
  mature(http//www.w3.org/TR/rdf-schema/)

19
Ontology layer - technology

• More metainformation, such as
• Transitive property.
• Unique, Unambiguous, Cardinality, etc.
• Ontology community exists- DL, OIL, SHOE, etc.
  etc.
• Not Turing complete, tractable.
• Like UML but different.

20
Ontology layer - goals and status

• Goals
• More powerful schemas (ontologies) allow
  inference and consistency checking.
• Decidable consistency.
• Equivalence links allow evolution of standards.
• Conversion and cross-reference between
  vocabularies.
• "Web of meaning.
• Status
• DAML (US) and OIL (EU) developments harmonized.
• Standards process well underway (OWL a W3C WD -
  2003/1) (http//www.w3.org/TR/owl-ref/).

21
Rules Layer - technology

• Adds variables to RDF.
• General purpose rules languages that allow query
  and filtering.
• Query similar to SQL.
• Rules layer allows form of Proof without full
  Logic layer.
• Monotonic rules essential, non-mon for closed
  world use.

22
Rules layer - Goals and status

• Goals
• Powerful ways of expressing relationships.
• Create new applications from scratch using
  webized rule engine technology.
• Equivalent to query language?
• Status
• Many rules systems currently exist, need
  webizing.
• Some common formats eg RuleML.
• Logical next step for a web standard.

23
Logic framework - goals and status

• Framework for writing axioms of rule-based
  systems.
• Monotonic logic.
• Any rule system can export, generally cannot
  import.
• No one standard engine - inference capabilities
  differ.
• Exchange proofs c between existing engines (SQL
  to KIF, Cycl, etc).
• Any system can validate proofs.
• Web assumptions different from closed world.
• Status
• Much academic discussion.
• Not ripe for standardization yet.
• Research agenda.

24
Web of Trust need

• All statements on the Web occur in some context.
• Applications need this context in order to
  evaluate the trustworthiness of the statements.
• The machinery of the SW does not assert that all
  statements found on the Web are "true".
• Trustworthiness is evaluated by each application.
• Very flexible language can express existing
  systems.

25
Web of Trust - goals and status

• Goals
• Small trusted codebase proof validator plus
  signature validator.
• Expressive power allows real trust structure to
  be expressed.
• Simple system allows security verification
  (Privacy, confidentiality, etc).
• Status
• Existing research demonstrates feasibility eg PCA
  (Felten, Appel). (http//www.cs.princeton.edu/sip/
  projects/pca/)
• cwm simple integration of cryptography and
  inference. (http//www.w3.org/2000/10/swap/)
• Some base standards exist, eg. XML Digital
  Signature.
• Large research agenda.

26
Conclusion of part 1

• Science and engineering and commerce will benefit
  enormously from the semantic web.
• We need research, standardization and deployment
  to ensure it happens.
• We can look to new research opportunities when it
  is here.

27
2. Some things we have been doing at MIT/LCS

• N3 language extends RDF up through the layers.
• Generic command line semantic web processors.
• Import and Export hacks for legacy systems.
• Generic database export.
• Native RDF applications.
• Prototyping the layer cake.

28
Application Integration Native RDF

• Teleconference schedule, present in bridge,
  action items etc (Swick's Zakim).
• Annotea is RDF-based web annotation
  system(http//www.w3.org/2001/Annotea/).
• W3C roadmap (by hand).
• RSS syndication feeds.
• Other project's metadata (e.g. RSS, Dublin Core,
  Adobe XMP, TAP,...).

29
Application Integration Import
Things we played with at MIT/LCS
30
Application integration Export
31
Application integration Case study Roadmap
http//www.w3.org/2001/04/roadmap/about.svg
32
Application integration Case Study Trip
http//www.w3.org/2002/08dc-ymx/make.svg
33
Case study - take away

• Runs across previous application boundaries.
• Involves personal, group and public information.
  (How do we prove it meets confidentiality
  requirements?)
• Many different data stores.

34
Web of Trust - simplified example
http//www.w3.org/2000/10/swap/test/crypto/make.sv
g
35
Conclusion to part 2

• The entire layer cake has been prototyped.
• We will have to bootstrap s/web with legacy data
  for some time.
• Lots of exciting opportunities in personal,
  enterprise global data.

36
3. Challenges for Research

• The Semantic Web Wave.
• Semantic Web bus - and above.
• What makes it a semantic web project?
• Challenges - Computer Science.
• User interface graphs to Graphics.
• Logic.
• Engineering choices.
• Conclusion of part 3.

37
The Semantic Web Wave
38
Semantic Web bus - and above
39
What makes it a semantic web project?

• Uses URIs for identifiers.
• Lookup of terms on the web by URI!
• Aware of information being from different
  sources.
• Properties of object not known in advance.
• Delegation to other engines.
• Based on standards (HTTP, RDF, OWL, etc...).

40
Challenges - Computer Science

• Indexing rule files by terms used -- input and
  output vocabularies.
• Using that index to resolve a query.
• Using it to show what cannot be deduced
  (privacy...).
• Building data flow and queries dynamically.
• Scalable algorithms over space with structure on
  all levels.
• Using same rules in eg. forward and
  backward-chaining contexts.
• Incremental, reversible inference - diff, patch
  and generic synch.
• Generating and checking lists of rules used
  proofs.
• Secure systems.
• User interfaces to the Semantic Web.

41
User interface graphs to Graphics

• Large array of UI metaphors.
• Outline views, forms, circles arrows, forms,
  charts, etc.
• Pretend there are trees.
• Pick up display style from style sheet - control
  by
• vocabulary author.
• data publisher.
• Reader.
• Manipulate overlays by data source, vocabulary.
• Keep the interface 2-way.

42
Logic

• A universal logic - language? framework?
• Overlapping model theories.
• In general, "layering".

43
Engineering choices

• Choosing size of dataset - cleanliness vs reach.
• When to query and when to download.
• Granularity of proof language.
• Legacy systems without proper models.

44
Conclusion of part 3

• The path to the Semantic Web needs foresight and
  encouragement.
• The opportunities from then on will be very
  diverse.

45
Thank You

• Questions/Discussion.

46
More information

• Semantic Web Home Page http//www.w3.org/2001/sw/
  
• Semantic Web Advanced Development Home Page
  http//www.w3.org/2000/01/sw/
• RDF Home Page http//www.w3.org/rdf/
• Semantic Web / RDF Interest Group
  http//www.w3.org/RDF/Interest
• Semantic Web / RDF IRC irc//openprojects.net/rd
  fig

47
Slides not used

• (If you reach this point you have gone too far).

48
Where W3C standards are

• Enabling Standards and Technologies for
  supporting the Architecture.
• The RDF Core Working Group http//www.w3.org/2001
  /sw/RDFCore/
• The Web Ontology Working Group
  http//www.w3.org/2001/sw/WebOnt/
• Semantic Web / RDF Interest Group.
• RDF Rules / Query - becoming WG soon?
• RDF Logic.

49
Web Services and Semantic Web

• Exchanging RDF messages more interesting than
  just XML.
• RDF Describing services.
• DAML Services http//www.daml.org/services/
• Challenge Automatic service composition.

50
1. Background and Principles
Blank Slide
51
The Semantic Web
Information Management A Proposal, Tim
Berners-Lee, CERN, March 1989, May 1990,
http//www.w3.org/History/1989/proposal.html
52
Goal building machinery

• "The bane of my existence is doing things that I
  know the computer could do for me."
• -- Dan Connolly, The XML Revolution.

53
Semantic Web Principles

• Any thing can have a URIxxx.
• Vocabularies can merge and be replaced with time.
• Documents are self-describing.
• "Anyone can say anything about anything.
• No one system knows everything.
• Design must be minimalist.

54
Do NOT wait for

• ...Artificial Intelligence itself.
• ...common sense reasoning.
• ...your gradmother to write in DAML.
• ...people to mark up web pages.
• ...your company to build the entire Semantic Web.

55
Short to medium term for s/w s/w developers

• Import Export data in RDF.
• Leverage data in existing applications.
• Web-enable all tools - web as file system.
• Identify low-hanging fruit in app overlap.
• Identify your niche - how to interface to
  partners.
• Co-develop rules with standard.
• Help W3C work out next steps.