here - University of Sydney

arghtalentΔιαχείριση Δεδομένων

31 Ιαν 2013 (πριν από 4 χρόνια και 7 μήνες)

198 εμφανίσεις

The Semantic Web,

Applications and Migration
Path


at HP Laboratories

Bernard Burg

Manager, Associative Metadata
Department, HP Labs Palo Alto

bernard.burg@hp.com


11 August, 2003

Sydney University

page
2

8/11/2003

Sydney University, Semantic Web & HP

HP Fast Facts



Company name: Hewlett
-
Packard Company



Headquarters: Palo Alto, California



CEO and Chairman: Carleton S. (Carly) Fiorina



HP serves more than one billion customers in more than 160
countries on five continents


141, 000 Employees Worldwide


$72B company

page
3

8/11/2003

Sydney University, Semantic Web & HP

HP’s Mission


HP's mission is to invent technologies and services that drive
business value, create social benefit and improve the lives of
customers

with a focus on affecting the greatest number of
people possible.



HP dedicates $4 billion (U.S.) annually to its research and
development of products, solutions and new technologies.



page
4

8/11/2003

Sydney University, Semantic Web & HP

HP is

#1 globally in personal computers

#1 globally in imaging and printing

#1 globally in enterprise storage

#1 globally in management software

#1 globally in UNIX, Windows and Linux servers

#3 globally in IT services

page
5

8/11/2003

Sydney University, Semantic Web & HP

HP’s 4 Global Business Units

Personal
System
Group

Enterprise
Systems
Groups

Imaging

and Printing
Group

Printing and
multifunction

Digital
Photography

Scanners
and
projectors

Supplies and
accessories

Servers

Storage

Networking

Utility Data
Centers

Adaptive
Enterprise

Desktops
and
workstations

Notebooks
and Tablet
PCS

Handheld
Devices

HP Services


Software
Services

page
6

8/11/2003

Sydney University, Semantic Web & HP

HP Labs’ roles


Contribute to HP strategy creation and alignment


Deliver technology that enables HP to win in HP’s
selected strategies through:


Breakthrough technologies


Technology advancements


Create new opportunities for HP that go beyond current
strategies


Invest in fundamental science and technology in areas
of interest to HP

page
7

8/11/2003

Sydney University, Semantic Web & HP

HP Labs Worldwide


Director & SVP Dick Lampman


~750 employees worldwide


~5% of $4B HP R&D budget

bristol

japan

israel

palo alto

cambridge

india

page
8

8/11/2003

Sydney University, Semantic Web & HP

HP Labs Major Research Areas

Solutions and

Services


Internet
Computing

Platform

Printing, Imaging

and Storage


Printing
technologies

Imaging
Technologies

Digital
Photography

Storage (MRAM,
ARS)

Utility Data
Center

Adaptive
Enterprise


Mobile Systems

Trusted Systems

Digital Media
Systems

Intelligent
Enterprise

Technologies

Semantic Web

Systems
Research

Data
Management

Advanced
Studies



Quantum
Information
Processing

Computational
Bioscience


Motivation for the Semantic Web


Application Integration


Market and Early Adopters


page
10

8/11/2003

Sydney University, Semantic Web & HP

Interaction Mediation

Shared Model at Interface (Ontology)

Buyer

Seller

Interface

Interaction

Buyers
Model
(Ontology)

Sellers Model
(Ontology)

Placed

Shipped

Delivered

Accepted

Cancelled

Returned

Order State

addToBasket

checkOut

Interaction
-
> effect in terms of
state change

‘Spontaneous’ state change
-
>
effect in terms of interaction

page
11

8/11/2003

Sydney University, Semantic Web & HP

Motivating the Semantic Web

from human<
-
>computer interoperability


to computer<
-
>computer interoperability


XML, the current web, is not adequate


captures structure, not semantics (relationships, constraints)


tags (properties) have no description


requires humans intermediaries to define mappings

RDF/OWL, the Semantic Web, has promise


RDF models objects, relationships


relationships (properties) are objects (have descriptions)


OWL adds rich constraints


machines can infer mappings (the big hope for interoperability)

The Semantic Web is a W3C standard

page
12

8/11/2003

Sydney University, Semantic Web & HP

The Essence of the Semantic Web


Semantic modeling

is key technology, because it allows
machine processing of metadata descriptions


XML lacks modeling power



Application integration today

RosettaNet

Appli x





Appli y

Msg 1






Msg 1

Msg 2






Msg 2










Msg k






Msg n

Manual mapping


-

Brittle


must
be updated as
msgs evolve

Maps elements,
not relationships


page
13

8/11/2003

Sydney University, Semantic Web & HP

Added Value of Semantic Modeling

Application integration with Semantic Web

Appli x





Appli y

Msg schema





Msg schema

ontologies

SW Compiler

Decrease:

-

manual mapping

-

integration time


Transformatio
n

Program

aggregation

Increase:

-

flexibility

-

resilience

Leverage
Semantic
Modeling







Msg 1






Msg 1

Msg 2






Msg 2










Msg k






Msg n

Msg l

page
14

8/11/2003

Sydney University, Semantic Web & HP

Potential Markets & Early adopters



Existing market:


Application Integration Software Market



$4.28B in 2001 $15.53B in 2006 (IDC #27236, June 02)



New spaces (for HP)


Adaptive Enterprise


The path to semi
-
structured data mgmt


SIMILE (HP, MIT, W3C) and Digital Publishing



Early adopters:


Adobe, Boeing, HP, IBM, SUN Microsystems, …



From the Semantic Web Standard
to

a Migration Path


page
16

8/11/2003

Sydney University, Semantic Web & HP

What’s the Technology Angle?

Why not just XML?

Unicode

URI

XML + NS + xmlschema

RDF + rdfschema

Ontology Vocabulary

Logic

Proof

Trust

Digital Signature

Self
-
desc.
doc.


Data


Data


Rules


Goal is
semantic

interoperability


XML gives data exchange standard
for consenting parties

Hard to reuse, hard to extend
schema, hard to merge data


RDF gives common data model
-
>
syntactic interoperation


URIs gives common space of
identifiers


Ontology layer gives explicit
conceptual model behind the terms


allows translation between schemas,
data integration, reuse, full
interoperation


Logic/proof layer allows exchange of
evidence chains (“believe this
because …”)


page
17

8/11/2003

Sydney University, Semantic Web & HP

Semantic Web Technologies


composable
, extensible fact/metadata representation

gives syntactic interoperability


RDF triple
-
model


representation for structure and nature of terms


ontologies

also composable and extensible

provides foundation for semantic interoperability


description logics


OIL, DAML, DAML+OIL, OWL (lite, DL, full)


techniques for translating between ontologies, ontology
-
based
data integration


proof and trust

layers for exchange of evidence chains (believe
because …)


page
18

8/11/2003

Sydney University, Semantic Web & HP

RDF in a nutshell


A data model for assertions about things labelled by URIs

plus an XML syntax


all facts are “subject
-
> predicate
-
> object” triples

these form a graph of assertions


predicates disambiguated by XML namespace


everything is a resource (or a literal string)


can “reify” assertions so can assert

“W3C
claims

‘RDF
importance

veryHigh’”

http://doc

Joe Smith

Illustrator

dc:creator

rdf:value

dcq:creatorType

page
19

8/11/2003

Sydney University, Semantic Web & HP

Ontology in a nutshell


a formal, explicit specification of a shared conceptual
model (aka domain model)


describes the terms used and their relationships


concept names and concept hierarchy


roles (predicates) and role hierarchy


concept expressions, associated axioms


could think of it as a glorified schema

entity
-
relationship models are a subset


page
20

8/11/2003

Sydney University, Semantic Web & HP

What HP does for you!

Nuin: agent toolkit

Joseki:

RDF
Canonicalisation

Source of picture: W3C

Jena 1

2002 Jena1 is used by more than 60% of the community

2003 Jena2 is still the leader

Jena2

page
21

8/11/2003

Sydney University, Semantic Web & HP

Progress on Technology




From Jena 1 to Jena 2

The RDF API

ARP

RDF Filter

n
-
triples

Readers

XML

XML

n
-
triples

Writers

Mem

RDB

BDB

Stores

The DAML
API

RDQL

Applications

Jena 1 Architecture

Jena 2 = Jena 1 plus



Full support for RDF2003

The RDF 2003 API



Generalised ontology API
with profiles for the OWLs,
DAML+OIL, RDFS,…

Ontology API

OWL
Full

OWL

DL

OWL
Lite

DAML

+OIL

RDFS



Event handling

Rule Systems

RDQL

Applications



Rule Systems



OWL Syntax Checker

OWL



Extensible reasoning support
for RDFS and OWL Lite,
including support for external
plug
-
in reasoners

RDFS

OWL Lite

External
reasoners



Fast
-
path database query (via
Genesis)

Fast path query



Efficient reification



Necessitated complete rearchitecture

Reification

Event handling

The RDF 2003 API

Jena SPI

Jena 2 Architecture

page
22

8/11/2003

Sydney University, Semantic Web & HP

Joseki

Coarse
-
grained RDF processing


A NetAPI


standard operations on models


coarse
-
grain wrapper to local fine
-
grained interaction


Application framework for RDF applications


Application paradigm


Publishing RDF data


Large RDF models


Multiple applications collaborating


Shared, updated RDF


page
23

8/11/2003

Sydney University, Semantic Web & HP

The Nuin agent engine


A
BDI agent engine
, written in Java


based on Rao’s AgentSpeak(L) languge [Rao 1997]


extensible knowledge representation based on FOPC with actions



designed to be programmer extensible at all points



default capabilities to make it easy to write agent behaviour “out
-
of
-
the
-
box”


interpreter for abstract actions


built
-
in action library of core capabilities


script parser for human
-
readable script syntax


abstract services to allow pluggable connection to infrastructure
service providers (FIPA platforms, SOAP, Joseki, etc)


integrated with Jena for semantic web processing


page
24

8/11/2003

Sydney University, Semantic Web & HP

Nuin architecture overview

agent core

reasoner

reasoner

reasoner

knowledge source

interpreter

knowledge source’

beliefs

desires

intentions

plan library

abstract service adapter layer

message service

directory service

Java object

invocation

JADE agent

platform

event and

message queue

evaluation

functions

action library

agent
configuration
(rdf)

RSS translator

example


concrete services

serialized
agent script(s)


page
25

8/11/2003

Sydney University, Semantic Web & HP

Genesis: migration path of SW



Optimization

Semi
-
autonomous
performance analysis,
benchmark, tuning,

STOR


Apps
(EAI/SIMILE)


Genesis


Jena 2


Database

Ease of Use

Higher
-
level objects

Immutability

Security

Distribution/Caching

Scalability

Support Jena 2 API

App
-
specific schemas

MySQL,

PostgreSQL,

Oracle


Efficient Graph Queries


Distributed Genesis models on top of scalable, persistent
Jena

page
26

8/11/2003

Sydney University, Semantic Web & HP

How to use these tools



Currently Available



Jena 2, Joseki


Open source
, BSD license, no restrictions to use


http://www.hpl.hp.com/semweb/index.html



Announced for October


Nuin


Open source
, BSD license, no restrictions to use


http://www.hpl.hp.com/semweb/index.html



Mobile Users,

Contexts,

Agents


page
28

8/11/2003

Sydney University, Semantic Web & HP






are mobile (small devices)



want immediate solutions to
their problems



have little time to waste


web browsing shows its limits
in this context

Our customers

page
29

8/11/2003

Sydney University, Semantic Web & HP





Topology of Mobile Applications

food

@home

Work policies

me

@HPL
-
PA

hypothesis


open world: any service, any object, real/virtual

micro
-
worlds mapped to


domains (inside firewalls, security, trust,
games, work)


cells (location awareness, cellular nets)


Need of policies and contexts


Need of semantic descriptions



Need of proactive behaviour

page
30

8/11/2003

Sydney University, Semantic Web & HP





Building Blocs: Standards



W3C Semantic Web (ongoing effort)



FIPA Agents solved the communication problem
between Agents (about 20 implementations, 7
open
-
source)



Agents are ubiquitous, from server, to laptop, PDA
and phone.




page
31

8/11/2003

Sydney University, Semantic Web & HP





Building Blocs: Large Scale Deployment

Paris


Dublin


Ipswich


London


Chambery


Lisbon


Barcelona


Parma


Saarbruecken


Berlin


Lausanne


Sendai


San Francisco


Sydney


Melbourne


Dunedin


Palo Alto


Salt Lake City


Honolulu


Miami


Open testbed

70 platforms deployed over 5 continents,

Agentcities Workshop at AAMAS

Web Services, work in Technical
committee with:

IBM, HP, Intel, Fujitsu, SAP, Sun
Microsystems, Mitre, Motorola…

page
32

8/11/2003

Sydney University, Semantic Web & HP





did we bring the solutions to
their problems?



less browsing

more active & pertinent
services


due to:

superior context awareness

agents proactivity


Added value to our customers


Next Steps


page
34

8/11/2003

Sydney University, Semantic Web & HP

HP as a research partner




HP is supporting several collaborations in
Australia (7?)



I am working on projects around mobility and
context


Monash U. Melbourne (2 projects)


Flinders U. Adelaide


Sydney U.






page
35

8/11/2003

Sydney University, Semantic Web & HP

HP
-

Australia in General




HP has high profile collaborations in USA:


MIT, Berkeley (Citrus project) …



HP has high profile collaborations in EU:


HPL Bristol has European projects, Universities,
student exchange…



HP has high profile collaborations in India:


HPL India has build an amazing network of
connections



In Australia, we should become the lighthouse
project for Semantic Web in context modelling to
become HP’s champions








page
36

8/11/2003

Sydney University, Semantic Web & HP

Raising HP’s investments in Australia is



Build a HP’s network of influence in Australia


Excellence in academia


Champions in “relevant” domains


Succeed in existing project


Establish student exchange


Link with industrial tissue


Shared projects with: HP, Academia and industrial partners


Develop market specific relations


Increase HP’s revenues Australia