Agents on the Semantic Web a roadmap to the future

drillchinchillaInternet και Εφαρμογές Web

21 Οκτ 2013 (πριν από 3 χρόνια και 7 μήνες)

62 εμφανίσεις

Agents on the Semantic Web



a roadmap
to the future

An arial view from 50 000 feet



Erik Lillevold, UniK

erlille@unik.no


Preface

In order to work systematically towards my PhD I want to use a method where I star
t from a
very high level of arial view, 50

000 feet, and step
-
by
-
step, come closer to the target. The
main purpose of this is to use this as a help facility to discuss my work with my supervisors
and others who are interested in my work
.

Introduction


The
initial view is just a little more text added to a PowerPoint presentation that I made to sort
out my first ideas of what I wanted to do.


Background
knowledge and literature

In the last years my interests and involvements
have been Web Services, SOA, Sema
ntic Web
and Semantic Web Services. However, back in 1990
-
ies I were involved in development of
technology called “Agents” that I and many others thought would be used in big scale in near
future. However, the Agent technology never

took off and is still j
ust played with in academic
demonstrators even though Agents is foreseen as a concept both in Semantic Web and
Semantic Web Services.


Following background literature will be used initially in my work:


o

Spinning the Semantic web by J. Hendler & all, MIT Pr
ess Ltd, 2003
-
03
-
04

o

Agents and the Semantic Web, James Hendler, University of Maryland


IEEE
Inteligent Systems, March/April 2001

o

Semantic Web Road map, Tim Burnerners
-
Lee, 1998
-
10
-
14

o

Semantic Web @ 5


Current Status and Future Promise of the Semantic We
b, James
Hendler/UoM and Ora Lassila/Nokia, Semantic Technology Conference, March 6
-
9
2006 San Jose, California

o

Agent Technology Roadmap
-

A Roadmap for Agent Based Computing, AgentLink
2005


Motivations

The
major impetus for my work is that
Semantic Web
f
inally
is about to leave pure research
and slowly picked up by industry
. However, even though
most of the fundamental technology
seems to be there

it is still necessary with m
ore research
and development of toolkits and
ontologies
.

Another shortage is inte
gration of technologies already developed within
Semantic
Web, Semantic Web Services and Agents
. My thesis will try to find a
roadmap for an
integration of those three technologies

through analysis of “where are we to day” and “where
are we going”. The dev
elopment of all the three technologies will be described in three stages:
Today (short term), Tomorrow (medium term) and Future
(long term).

Today


Semantic Web

In the research and development is the l
ower layers
of the proposed Semantic Web stack quite

m
ature in the context of research, i.e. XML, RDF, Ontology,
etc. However, it is still a
”hen
and
the
egg
” problem when it comes practical usage in real
-
world applications. Especially are
commercial tools and
free open source software of good quality more o
r less missing in order
to get the ball rolling.




Figure
1

Source: Semantic Web @ 5


Current Status and Future Promise of the Semantic Web,
James
Hendler/UoM and Ora Lassila/Nokia, Semantic Technology
Conference, March 6
-
9 2006 San Jose,
California


In
Figure
2

is shown several applications made by tools and ontologies developed in the
project MINDSWAP at University of Maryland. It demonstrates that even simple ontlogies
and t
ools cane make quite advanced applications. The lesson to learn is that it not necessary to
wait to start making Semantic Web applications
.



Figure
2

Example applications of Semantic Web


Semantic Web Se
rvice (SWS)

The SWS has taken most of its concepts from Web Service and Service Oriented Arch
itecture
(SOA), but have the concept of machine readability, i.e. ontology description of service, in
common with Semantic Web. Today there are three major resear
ch and standardisation work
going on in parallel, i.e.
SWSF, WSDL
-
S, OWL
-
S and WSMO. A brief description of each of
them are found in
















Table
1

Overview of research and standards initiatives today


Agents

We know that a lot of research has been done and still does. They have been foreseen as
important components in Semantic Web, but many ask the question “Where are the agents
?

a
nd “Will they ever be used
outside academia?” The history of development of Agents is
long, least since early 1990
-
ies. First the major work was done in USA
, in military and
academic environment with some industrial attempts. In the last years has the heg
emony been
more and more been taken over by Europe
an organisations like

the AgentLink,

Agentcities
and FIPA.

AgentLink is an activity partly supported by EU to coordinate all Agent projects running in
EU FP program. One of the greatest contributions from A
gentLink is their Agents roadmap
that is the continuously updated
.



2005/6
Standards

WSMO

SWSF

WSDL
-
S

OWL
-
S

Description

WSMO (Web
Service Modeling
Ontology) has been
developed

since
2002. Part of a larger
framework (together
with Web Services
Modeling Language
and Web Services
Execution
Environment)

SWSF (Semantic
Web Services
Framework) is an
effort by Semantic
Web Services
Initiative (SWSI).
The first meeting of
SWSI was held

in
December 2002.
SWSF uses DAML
-
S draft as a starting
point.

WSDL
-
S is being developed in
METEOR
-
S project. The main
purpose of WSDL
-
S is to extend
existing WSDL standard with
semantic description.

OWL
-
S is an

OWL
-
S is a OWL
-
based
Web service
ontology
and

research effort
started in May 2001
as DAML
-
S. In
November, 2003 the
language was
renamed to OWL
-
S.
Current specification
1.1 dates to
November, 2004.

Developers

Digital Enterprise
Institute (DERI)

Language Committee
of
SWSI

IBM and University of Georgia.

Many companies and
universities working
in DAML program.

Standardization

Submitted to W3C
June 3, 2005 [
link
]

Submitted to W3C
Septe
mber 9, 2005
[
link
]

Plans for W3C submission.

Submitted to W3C
November 22, 2004
[
link
]

More info

http://www.wsmo.org

http://www.daml.org
/

services/swsf/

http://lsdis.cs.uga.edu
/projects/meteor
-
s/wsdl
-
s/

http://www.daml.org/

services/owl
-
s/







Figure
3

Agent technology comprises areas that will be addressed over different timescales


Even though FIPA is located in Switzerland it as large number of memb
ers from all over the
world and is in practice the number one standardisation body within agent technology. More
than 50 larger industrial companies have supporting their work on standards since 2002. A
large number of standards are at the moment quite mat
ure and stable. Their international
acceptance is supposed to increase after the further

work on agent technology newly was
taken over by IEEE. For several years now it has been developed both tools and
demonstrative applications based on FIPA standards.


Agentcities is probably the biggest and demo where FIPA standards were used to make
experimental applications on a network of agent platforms. Due to lack further financial
support is this experiment no closed down.


The far most popular
toolkit for the FI
PA standards is JADE (Java Agents Development
Environment). It is mature and heavily tested. It has now reached industrial quality and is
ready to be used in commercial systems. The upper architecture of JADE is depicted in



F
igure
4

The JADE architecture

Tomorrow and
further
Future

When moving from today where we know what conditions are in the three disciplines of
technology, we very much have
identify development trends to watch what innovative
resea
rch and industry think about the future of Semantic Web, Semantic Web and Agent
technology.


Semantic Web

In the period of 2001
-
2002 had Jim Hendler and Ora Lassila views of the Semantic Web
applicatition that are respectively depicted in
Figure
5

and
Figure
6
.


Most things predicted have happened or are happening at the moment.
Som
e things happened
faster than they

anticipated
, e.g.
triple store scaling
,
reasoner performance actually matters

ad
ontologies are t
here
in spite of very little linking
. However, s
ome
other
things are yet to
be
materialize
d, such as:

• public information sources
based on RDF, OWL, etc

• digital convergence and

pervasive computing
is
emerging
, but is not just at right
stage

• progress

on agents

is not
as expected


Now

the
y

believe the time has come

to
go out there and
build

real
-
life application
s of what
already is

in place and
hopefully
make
some money
out
of it.



Figure
5

Jim Hendler’s view of Semantic Web
in 2000
-
2001




Figure
6

Ora Lassila’s view of Semantic Web in 2000
-
2001


Semantic Web Services

Semantic Web Services are closely related to SOA. The difference is mostly that the former is
generic and include all kind of applicati
ons, while SOA more goes in the direction of business
applications. Another difference is that a SOA system does not emphasis on
automatic real
-
time generation of services, while Semantic Web Services does.

The work on SOA will
mainly be driven by major so
ftware vendors like IBM and Microsoft, while Semantic Web
Services will steadily be made more mature by cooperation between academic and industrial
research departments. W3C will play an important role as standardisation forum for SWS.


The projects mentio
ned in
Table
1

have already made their first demonstrations based on their
respective
“standards” and service lifecycles (se fig





Figure
7

Example lifecycles of prototyped Sem
antic Web Services


It is expected that the different initiatives will gradually harmonize towards one common set
of standards for Semantic Web Services under supervision of W3C.


Agents

o

FIPA

o

IEEE takes over FIPA

o

JADE is becoming increasingly improved with

a range of toolkits to build applications

o

It does not seems to be much real competition to FIPA and JADE

o

Agentcities was laid down due to loss of support from EU IST

o

AgentLink (III) is still supported financially by EU IST.


Integration roadmap

There are

several signs that the technologies described above are starting to merge. At DERI
in Austria several interesting projects are going on:

o

Triple Space Computing (TSC)
.




-

TSC is new communication and coordination paradigm for machine
-
machine
interaction.

-

T
SC project will be a

proof of concept through prototype

implementation and use
case analysis with Semantic

Web service.

-

Technol
ogy approach is

re
-
arrangement and combination of

established Web
technology and Shared Object Space

middleware.


The overall arc
hitecture is seen on




Figure
8

TSC overall architecture


o

DERI project INFRAWEBS (
Intelligent Framework for Generating Open (Adaptable)
Development Platforms for Web
-
Service Enabled Applications Using Se
mantic Web
Technologies, Distributed Decision Support Units and Multi
-
Agent Systems
):

-

The overall architecture is shown in


o

DERI project SEnSE (
Semantic Engineering Support Environment
).
SEnSE will
provide an environment where
:

-

Users can delegate change t
racking tasks to software agents that proactively
monitor the evolution and changes of design artefacts (tools)

-

Notifications can be based on the semantic structure of documents / resources
rather than simple version changes

-

Information about changes can b
e done for indirectly relevant artefacts (tools) as
well























Figure
9

INFRAWEB’s
Service Access Middleware (SAM) architecture


Conclusions

The state of the art for Semantic Web, Semantic Web Services and Agent tec
hnology is
briefly described. Further, the future trends and activities are indicated. They show that the
process of integrating those three technologies has started and will continue in considerable
pace towards t
he next generation web.