Semantic Web Design

convertingtownΛογισμικό & κατασκευή λογ/κού

4 Νοε 2013 (πριν από 3 χρόνια και 11 μήνες)

129 εμφανίσεις

INFS1300
-

Semantic Web Design

Andrew McInnes

41736731

Semantic Web

Design

Extracting Meaning
from

Web Content


The Semantic Web is a vision by the World Wide Web
Consortium (W3C) where data on the WWW is
converted into a common format
, known as the Resource Description Framework (RDF) so that this
data may be read and interpreted by a machine. This would allow data to be shared and reused
more easily between applications without a human user
having to input or manipulate the data
manua
lly.
This means

tha
t the computer could

do more
of the
mundane tasks such as locating,
combining and

presenting information on the W
eb.
But is this Semantic Web possible

in this day and
age, or is it still just a futuristic dream
.

Tim Berners Lee first in
troduced the idea of a Semantic Web back in 2001 went he presented the
notion of a ‘web agent’
(a web server, web service,
or database)

which, in this case, was scheduling
medical appointments using information such as
user
calendar
entries
and
the doctor
’s
appointment
times to calculate appointments that would not clash. Then finding the medical practise closest to
the user’s home address and scheduling the appointment. All the user has to do is accept the
appointment, if for some reason they still aren’t

happy they can decline the appointment, in this
case they are presented with the next best option and so on until they have an appointment.

No
need to manually compa
re times or search through the W
eb for available doctors.

The W
eb of today
is still far fr
om Berners Lee’s vision. Although a number of recent advances may mean it is closer
than ever before.

Text Reading software has come a long way in recent years, in order for the Semantic Web to
become a reality it is important not just for machines to re
ad text, but to also understand what that
text means. As far back as 2006 systems were being developed to try and achieve this. One such
system is TextRunner, instead of being told what the relationship between c
ertain entities in a
sentence
, TextRunner le
arns the relationships by extracting the information and constructing textual
‘triples’. A triple is a set of three strings where the first and third denote an entity
, and the second
string is the relationship between them. Take the sentence “Steve hired B
ill to
paint his fence” a
triple that c
ould be extracted would be “Steve, hired, Bill”
w
ere “hired” is the relationship between
them both.

Another triple extractable from that phrase is “Bill, paint, fence”.

TextRu
nner can run
through an entire W
eb documen
t and store all of the triples it comes across. Of course this has its
problems such as entity duplication and inconsisten
cies between those entities,

these things will
have to be considered before text reading software can be considered
perfect.

Once you have the machine reading
and understanding text off the W
eb, you then have to consider
how they wi
ll communicate with each other. Hitzler and
Harmelen

explain in
A reasonable Semantic
Web
, that

tw
o agents
A and B
can only

be truly semantically int
eroperating

if
B can understand the
literal cont
ent of a phrase A is saying, as well as

understand the implicit relationships of that phrase.
This is why the RDF Schema is such an important component of the Semantic Web, without a
common data format, commu
nicating between machines would be near impossible (imagine trying
to talk to someone who is using a foreign language, you can try and understand but really you have
no idea what they are trying to say). Using the RDF
it is
even
possible for agent A to enf
orce beliefs
INFS1300
-

Semantic Web Design

Andrew McInnes

41736731

on the receiving
agent B by specifying domains and ranges of a property. An example would be if

A
transferred the

phrase “Canberra is a city”

to B. I
f
you change the phrase

to “Canberra is the capital
of Australia” the property “is capital of”

infers a
lower bound on B meaning that B has to accept that
Canberra is the capital of Australia and therefore no other city can be. These kinds of phrases help
form relationships without explicitly having to say them.

The next stage of the Semantic Web
, now that the mac
hines can read in text off the W
eb and
communicate it to difference machines and applications is how to then display this semantic
knowledge. The Semantic Search
Engine
is one such way that this information can be gathered and
presented.
Recently Wolfram Alpha has
gained popularity as one of the closest things to a Semantic
Search. Instead of a traditional search where you type in what you are searching for and you receive
pages and pages of possible matches that you then have to check, Wo
lfram Alpha asks what you
want to know, then displays its idea of an answer for that question. You can see how it comes to that
answer and how it interprets your question, which you can then alter to change the outcome of your
question.

Wolfram started off

as a simple calculator for mathematical equations, it has now grown
to include categories such as weather, geography,

history, music, and many more i
n an effort to be
truly semantic and be able to answer any question you need answering.

Unfortunately Wolf
ram is
not perfect and still requires some work in order to be considered a true Semantic Search Engine.

The final hurdle of the Semantic Web, once all the technology is there, is whether or not it can gain
the critical mass needed to become the web stand
ard of tomorrow. Hendler notes in his paper
The
Semantic Web
that an event similar to that of the Web’s early days must occur, people must become
invested in this idea that the Semantic Web will greatly benefit

them. Currently web users will not
mark up th
eir websites using the RDF because they see no immediate advantage in doing so. In turn
no applications will be created that can demonstrate these advantages, meaning the Semantic Web
will hit a dead end and go nowhere.
Another problem with creating conten
t for the Semantic Web is

making sure that it is simple enough that an average web user can create
machine
-
readable

code,
ideally without them even being aware of it. DARPA
, with close help from W3C and other institutes,

is currently working on a tool call
ed the DARPA Agent Markup Language (DAML) which aims to solv
e
these problems. B
y funding the research for the program and creating
their own
content using the
DAML
, they aim to encourage other web users t
o switch over to a Semantic Web design.


A
ll these advances in technology once combined will form the basis of the Sematic Web. In the
future we may be able to book a flight to Melbourne, along with a hire car that will be waiting
for
you
when the plane lands, to take you to a hotel room that has
been booked using your credit
card
and automatically discounted

any reward programs you a
re

joined to,

all this

just by telling your
electronic device that you want to fly to Melbourne

and the day or date you wish to leave
.

The uses
for a Semantic
Web are
almost limitless, using tags that many social websites support to better
understand the content and enable more useful answers when searching for a topic
,
help with
medical

diagnosis and supporting doctors while managing patient treatments
,
use in
enterpri
se,

such as gathering content from different sources and applications and combining them into
meaningful data.

But until the technology can be perfected and integrated to work together, that
future will

still
be out of reach.

INFS1300
-

Semantic Web Design

Andrew McInnes

41736731


Bibliography


http://www.w3.
org/2001/sw/

http://semanticommunity.info/@api/deki/files/4686/=10.1.1.115.9584.pdf

http://turing.cs.washington.edu/papers/aaai06.pdf

http://iospress.metapress.com/content/9717070651132833/fulltext.html

http://www.wolframalpha.com/about.html

http://hcs.sci
ence.uva.nl/Capita
-
AI/2002/papers/hendler.pdf


ftp://ftp.ksl.stanford.edu/pub/KSL_Reports/KSL
-
00
-
10.html

http://www.altova.com/semantic_web.html

http://www.iospress.nl/journal/semantic
-
web/


http://www.w3.org/2005/04/swls/

http://iswc2004.semanticweb.org/posters/NI
-
Pollock
-
ISWC
-
Poster.pdf