SW status and direction

http://www.w3.org/2003/Talks/1023-iswc-tbl/

Tim Berners-Lee

MIT Computer Science & Artificial Intelligence Laboratory (CSAIL)

World Wide Web Consortium

International Semantic Web Conference 2003

This talk

• Progress
• Explaining/Evangelizing SW
• Fun stuff

Progress!

• OWL becomes stable
• Steadily growing deployment of RDF
• Growing SWeb-specific industry sector
• Varied acceptance in conventional IT
• SW Services starting to take off

Well done

Risks

• Architecture becomes fractured, weak, or baroque
• Fracture between Web and S/Web arch
• Fragmentation in query and rules
• Balancing research vs engineering
• Patents
• RDF/XML syntax shock
• Perceived relationships between SW and WS
• Timing and expectations
• Deployment in real products

Deployment

• Work RDF into products
• Legacy adapters - registry - shopping list
• Evangelism: using other people's terms
• Find low-hanging fruit

When you have a hammer

(power screwdriver and self-tapping screws?)

• Domain Name System
• IMAP
• WebDAV
• UDDI
• vCard
• iCalendar
• SyncML
• OFX
• CVS
• GIS! geographic information systems..
• Registries
• ...

Just use HTTP + RDF

The Killer App for the Semantic Web

• No such thing
• Its the integration, stupid!
• Look for 1 groups you might have overlap with

Semantic Web Evangelism

• Be careful of terms - ontology, semantics, etc
• Sell S/W on medium-term goals
• The three phases of adoption
• Aim at specific market - Enterprise, PIM, Government
• WS-SW relationship - orthogonal to 1st order
• Explaining how communities interact

SW misunderstandings

• - One big web - trust everything
• - One inconsistency trips it all up
• - One big ontology
• - AI has promised us so much before
• SW points to make
• - Communities of all sizes

Applications connected by concepts

2003: Enterprise Application Integration

RDF: Enterprise Integration hub

Global Integration bus

RDF is to enterprise software what IP was to networking - Keith Fox, Brandsoft

We must now deliver

the rest of the kit

- Specifically, remote query, rules.

Query? A basic web service

- with interesting properties, completely defined mathematically & no side-effects

- beneficial to describe in depth

Interlude: Web Architecture^(TiM)

Loosely,

• Identify things with URIs
• It is useful to be able to dereference them
• We all have to men the same thing by the same URI
• HTTP is space of information objects
• The # separates a global information object identifier from a local identifier in language in question
• Each system uses
• Basically hypertext and semantic web are distinct languages, systems
• For HTTP, internet media type indicates language if/when access happens

Lots of discussion - TAG, public-sw-meaning@w3.org, etc

Please re-use

• Don't create new URI schemes
• Don't re-invent HTTP space
• Don't re-invent RDF
• Don't re-invent ontologies where they exist

Weblike

• Test of Independent Invention
• Structure of all sizes
• Many structures underneath: RDB, XMLDB question?
• Dereferencing things is the way it works

When to look things up

• You never have to
• It is what puts the Web in Semantic Web
• "Ontological closure"
  • Look up predicates and types (cwm --closure=pt)
  • Nice for it to be O(1) or log(n) in the size of the semweb

  Look up predicates and types

• Finding definitive sources of data
  • Documents
  • Services
• Looking up rules
  • What sort?
  • What guarantees?
  • What result?

Simple example

name a city in a state that borders Massachusetts ?

{  ?m  state:code "MA".
   ?s  state:borderstate ?m.
   ?c  city:state  ?s.
   ?c  city:name ?n
} => {
   ?n a :Name_Of_A_City_In_A_State_Bordering_Massachusetts
}.

Terms on the semantic web.

@prefix state: <http://www.w3.org/.../data/USRegionState.n3#> .
@prefix city: <http://www.w3.org/.../data/USCity.n3#> .
{  ?m  state:code "MA".
   ?s  state:borderstate ?m.
   ?c  city:state  ?s.
   ?c  city:name ?n
} => {
   ?n a :Name_Of_A_City_In_A_State_Bordering_Massachusetts
}.

data/USCity

<> is log:definitiveDocument of :name, :state.

    :USCity     a rdfs:Class;
         rdfs:comment """
      Information about US cities.
    """;
         rdfs:label "USCity";
         rdfs:subClassOf city:City .
    
    :abilenetx     a :USCity;
         :name "Abilene";
         :state usstate:TX .
...    
    :albanyny     a :USCity;
         :name "Albany";
         :state usstate:NY .
    

cwm WhatCity.n3 --mode=rse --think

"Albany"     a :Name_Of_A_City_In_A_State_Bordering_Massachusetts . 

"Amherst"     a :Name_Of_A_City_In_A_State_Bordering_Massachusetts .

"Avon"     a :Name_Of_A_City_In_A_State_Bordering_Massachusetts .

"Bridgeport"     a :Name_Of_A_City_In_A_State_Bordering_Massachusetts .

"Buffalo"     a :Name_Of_A_City_In_A_State_Bordering_Massachusetts . 

"Burlington"     a :Name_Of_A_City_In_A_State_Bordering_Massachusetts .

"Concord"     a :Name_Of_A_City_In_A_State_Bordering_Massachusetts . 

"Danbury"     a :Name_Of_A_City_In_A_State_Bordering_Massachusetts . 
...  

Useful to be able to look things up

Looking things up

• When should my inference engine dereference a URI?
• What breadcrumbs should I leave for others to follow?
• What terms do we need to define to allow this?

Fun things coming up

imagine that there is a lot data...

Indexing data - by ontology

Indexing rules, building translation paths

Whatever happened to delegated database query?

like one big database? or one big web?

Indexing ontologies

Generic sematic web change propagation

• - Publish-subscribe model

  - Graph diff and patch

• - Synch for arbitrary information

The SW browser - RDF -> SVG via (Javascript? Rules? adrenaline?...)

How's the stack doing?

The Semantic Web Wave

Not to mention

Connections WS-SW

• Discovery should all be SWeb-based
  • But where is RDFQ and RDFQP?
• RDF as SOAP payload
• RDF Query plugs in wherever XML Query (or XPath) is used.
• SW Business rules engines call out to WS
• Remote RDF Query could use SOAP
• Remote RDF update should use SOAP.

Announcement

W3C Semantic Web Services Interest group now exists

What SWWS can offer WS

• Get rid of DTD/schema-fragility
• Design process for WS architecture
• Bottom-up goal/constraint based composition of WS

  Semantics of message content!

• Provable results
• Trusted systems

Trusted systems and SW

• Real SW applications need quotation, inspection
• Algorithms for solving problems will be very varied
• Proof language can be very simple
• Proof checker + digital signature verifier = secure agent
• Small trusted code base

Balances

• Engineering vs Research
• Getting it working vs getting it right

  Tractable Machinery vs Heuristics

When we have a vast array of data then things may look very different.

Let us make one

Thank you for your attention

http://www.w3.org/2003/Talks/1023-iswc-tbl/slide1-1.html

Web Services and Semantic Web

WS do program integration

SW does data integration

... across application barriers

... across organizational barriers