Introduction to-the-semantic-web

>oa 04-05- pec> O
.org/ 001/XM / http://
L
p>20 re</setS s
hive .org/2 ai_d
c
m tu narc .w3 /o
c >Pi c w.ope /www I/2.0
://ww "http:/ s.org/OA
r> d c= "http :xsi= hive :title
>
a ta> lns:oai_ 1/" xmln .openar d">
s c
e ]</dc
m . w s ictur
c
Web 3.0, an introduction to the
rg / L
l
o
1
:dc x ements/ http://ww /oai_dc. 32, 3] [p
dc/e cation=" /OAI/2.0 e, [ca. 1
rg d g
x

r
9
> cm.<
/dc:f
o rmat
>
d
uring -
ge d .pic
ch ema chives.o bour Bri ublishe .6 x 8.9 u r Bri au/nla
ar ar c:p p;w ; 6 rage> arbo a.gov.
w
>
n
S semantic web
.ope ydney H 1932]</d &am
.
:title sher>[ca tograph Wales</
:b d c:co
v e
e
of th rnet a
H
ney tp://nl
Syd t: ht
bli o
:pu at>1 ph w Sout verage h > o gr aphs e Inte b ject> t>
c
m e c o phot on via th c:su ubjec
d c:for erage>N 32</dc: u m of versi h s.</d /dc:s
o v 19 date> n : Alb ronic grap phs.< ction
dc:c verage> /dc: l ectio elect
to
Pho otogra o nstru
co 2< f col in an ey -- -- Ph ing c
<dc: ate>193 n>Part o lable - Sydn .W.) g e du
r
<d c:d criptio vai on> @paulhagon les - y, N.S r Brid
e s lso a cripti th Wa ydne rbou
dc:d ction.; A :des Sou ge (S y Ha
<
tru 8.</dc -- New r Brid yd n e it for
cons 829 d ges arbou f the S t
e
o us oll/
vn 307 ject>Bri ney H p hs o > wish /digic
u b y d e>< hotogra o urce y. If you ov.au
dc:s bject>S </dc:typ of p r > /dc:s stud .g o n.</
<
su ge um ntifie ustralia< h and w.nla ermissi
<dc: ype>Ima rt of: Alb d c:ide of A a rc tp ://ww uest p
c:t Pa > 98</ ibrary rese quot;ht to req
<d lation> elation 0782 nal L e for ref=& t;
dc :re ].</dc:r c -vn3 atio i mag lt;a h /a&g
e la.pi d by N this he & i a<
r
pictu entifier> m hel
n
r print tact t stral
[
id te ve o st con of Au tifier
>
<dc: ource>I may sa u mu ibrar
y
c :iden
c:s ts>You es, yo nal L 8</d
<d igh o s t;Nat
io
0 7829
< dc:r er purp quot;&g c -vn3
oth html& /n la.pi
any ight. v.au
yr
cop hts> n la.go
ttp://
d c:rig ntifier>h
<d c:ide

http://www4.wiwiss.fu-berlin.de/bizer/pub/lod-datasets_2009-03-05_colored.png
Semantic web, you probably see diagrams that make no sense like this.

PREDICATES
OWL
CURIE
TRIPLES
SPARQL
RDF
Lots & lots of confusing scary acronyms

Web 1.0
1993
Hyperlinks, static,
consumption

It’s worth going back in history. Web 1.0 was static. Publishers built things & people came & read them.

Web 2.0
2004
2 way, interactive,
producing, social

Web 2.0 - social web. People became the publishers, websites became the mechanism to allow people to connect (Flickr, Twitter,
Facebook, Wikipedia).

Web 3.0
~2010
Semantic, machine driven

Web 3.0 - semantic, machine to machine, putting meaning behind the structure/content of websites.

What does semantic web mean. Lets say you do a search for a recipe on Google. Notice the search results, there’s reviews with
star ratings, there is ingredient facets, cooking time facets, kilojoule facets. How does this work?

Microdata, Microformats and RDFa. All methods of describing content and the relationships between content.

<a href=”http://creativecommons.org/licenses/by/3.0/”>
<img src=”http://i.creativecommons.org/l/by/3.0/88x31.png”>
</a>

Let’s look at the basic HTML structure behind these mechanisms. Something every page has - license information. This is
typically marked up like this as a link. There’s no meaning to the link, it could be a link to anywhere.

<a rel=”license” href=”http://creativecommons.org/licenses/by/3.0/”>
<img src=”http://i.creativecommons.org/l/by/3.0/88x31.png”>
</a>

By adding in an attribute describing what the link is, suddenly we have turned the link from a normal link into something with
some semantic meaning. This is what all of these formats do, they add in extra descriptive markup.

schema.org A place that describes all the markup you can add for search engines (Google, Bing, Yahoo).

At this page you can see schemas for all sorts of common things - books, maps, music, photographs, recipies, reviews,
scultpure, organisations, places and more.

<div xmlns:v="http://rdf.data-vocabulary.org/#" typeof="v:Recipe">
<h1 property="v:name">Grandma's Holiday Apple Pie</h1>
<img src="apple-pie.jpg" rel="v:photo"/>
By <span property="v:author">Carol Smith</span>
Published: <span property="v:published" content="2009-11-05">November 5,
2009</span>
<span property="v:summary">This is my grandmother's apple pie recipe. I
like to add a dash of nutmeg.</span>
<span rel="v:Review">
<span typeof="v:Review-aggregate">
<span rel="v:rating">
<span typeof="v:Rating">
<span property="v:average">4.0</span> stars based on
<span property="v:count">35</span> reviews
</span>
</span>
</span>
</span>
Prep time: <span property="v:prepTime" content="PT30M">30 min</span>
Cook time: <span property="v:cookTime" content="PT1H">1 hour</span>
Total time: <span property="v:totalTime" content="PT1H30M">1 hour 30 min<
span>
Yield: <span property="v:yield">1 9" pie (8 servings)</span>
<span rel="v:nutrition">
<span typeof="v:Nutrition">
Serving size: <span property="v:servingSize">1 medium slice</span>
Calories per serving: <span property="v:calories">250</span>
Fat per serving: <span property="v:fat">12g</span>
Let’s look at </span> the Google recipe example. This is marked up with RDFa. Don’t worry too much about the meanings
the HTML of
</span>
yet as I’ll try to describe them all later.
Ingredients:

2009</span>
</span>
</span>
</span>
</span>
span>
This deﬁnes </span>information about the content of the page.
some basic
</span>
Ingredients:

2009</span>
</span>
</span>
</span>
</span>
span>
</span>
This describes the star ratings
</span>
Ingredients:

</span>
</span>
</span>
</span>
span>
</span>
</span>
Ingredients:
<span rel="v:ingredient">
<span typeof="v:Ingredient">
Thinly-sliced <span property="v:name">apples</span>:
<span property="v:amount">6 cups</span>
</span>
</span>
<span rel="v:ingredient">
<span typeof="v:Ingredient">
<span property="v:name">White sugar</span>:
<span property="v:amount">3/4 cup</span>
</span>
</span>
Further down it’s marked up with times, nutritional values, ingredients, etc.
...

So you can see from the markup, how Google can use this information to build the results page with the reviews, the facets for
time, ingredients and nutritional value. If those added bits of information weren’t added into the markup this couldn’t be built.
Pages that have this extra markup in them tend to rank higher in Google that pages that don’t have this speciﬁc markup in
them.

Let’s enter into the world of the semantic web. Now there’s lots of scary terms when you start heading down this path - Triples,
predicates, SPARQL. The basics really aren’t that scary, but immensely powerful. So let’s dive into the basics of RDFa.

R
Resource
D
Description
F
Framework
a
attributes

RDFa - RDF in attributes.

Frank Hurley

Lets start with a subject

Frank Hurley
takes

a predicate

Frank Hurley
takes
photographs

and an object.

Subject Frank Hurley
Predicate takes
Object photographs

These three things make up a triple. with a subject, predicate & object.

Subject

http://nla.gov.au/nla.pic-an25043941

You have a subject - in this case the URL of the photo

Objects
Exterior of.... -37.816068

144.961201

Museum

http://nla.gov.au/nla.pic-an25043941

1961 1 negative : b&w ; 12.5 x 10.0 cm.

You have objects that are related to the subject, but they are meaningless without...

Predicates
Exterior of.... -37.816068

144.961201
Title
Latitude
Museum
Longitude

Coverage http://nla.gov.au/nla.pic-an25043941

Date Format

1961 1 negative : b&w ; 12.5 x 10.0 cm.

Predicates. These form triples.

Finding namespaces

One of the beauties of this system is it’s supposed to be extensible, if there isn’t an existing namespace you can write your own
(yuk). You can guarantee that 15 different organisations will write 15 different versions. We’ve seen this already in the examples
from schema.org, Dublin Core

CURIE
dc:creator

http://purl.org/dc/elements/1.1/creator

The property called creator
from the vocabulary
identified by dc

Compact URI’s called Curies.

dc:creator
An entity primarily responsible for
making the resource.

Examples of a Creator include a
person, an organization, or a service.
Typically, the name of a Creator
should be used to indicate the entity.

It’s something that has a meaning & it’s something that should be being used consistently across all the cases that use it.

dc:title
dc:creator
dc:subject
dc:coverage
dc:relation
dc:identifier
dc:rights
dc:type
...
So we can ﬁll in the blanks for all of these bits of information

foaf:person
foaf:knows
foaf:isPrimaryTopicOf

In addition, we can also mark it up using other schemes, so people can be marked up with their relationships.

Let’s have a look at Flickr. With Flickr you can add tags, to help describe the image. If you add in tags in a special format they
are known as machine tags.

Now “machine tags” are a special type of tag. They actually start to describe the image & provide extra information. In this case
geo:lat= something geo:lon= something. However, they don’t actually link to a description, so although they copy, there isn’t
quite the same degree of semantic structure to them.

In the library world we typically have a MARC record as our master record. Very ugly, but very structured. Fields with all their
arcane subﬁelds, each is deﬁned and has a distinct meaning.

We sometimes transpose that data into other formats. This is an export of our data from our OAI-PMH interface & you can see
that it’s in XML & it’s rich, everything has a value, it conforms to a standard structure - Dublin Core. Compared to MARC it’s a bit
more human readable.

We then take this information and present it on the web in HTML. But if we look at a web page, this rich markup is lost. In our
HTML we have basic markup that Gives a bit of weight to various parts of the page.

H1

We have an H1 tag that gives some weight to the most important thing on the page - the title.

H1

TH TD

We mark up the rest of the information in a table. There’s an implied relationship between the table headings and the table data,
but we’re not saying what that relationship is. It could be anything.

At a basic machine level, there really isn’t a lot that is describing these elements and there isn’t anything that makes these
speciﬁc.

<div>
<h1>[Sydney Harbour scene with naval vessels in foreground, and Harbour Bridge in
distance] [picture] /</h1>
<table>
<tr>
<th>Date</th>
<td>[between 1910 and 1962]</td>
</tr>
<tr>
<th>Format</th>
<td>1 negative : acetate, b&w ; 8.7 x 12.1 cm.</td>
</tr>
<tr>
<th>Online Versions</th>
<td>http://nla.gov.au/nla.pic-an23417397</td>
</tr>
</table>
</div>

In this case it’s a table. there is an implicit relationship between the table header & the table data

<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML+RDFa 1.0//EN" "http://www.w3.org/MarkUp/DTD/
xhtml-rdfa-1.dtd">
<html xmlns="http://www.w3.org/1999/xhtml"
xmlns:dc="http://purl.org/dc/elements/1.1/"
xmlns:foaf="http://xmlns.com/foaf/0.1/">

<div about="http://nla.gov.au/nla.pic-an23417397">
<h1>[Sydney Harbour scene with naval vessels in foreground, and Harbour Bridge in
distance] [picture] /</h1>
<table>
<tr>
<th>Date</th>
<td>[between 1910 and 1962]</td>
</tr>
<tr>
<th>Format</th>
<td>1 negative : acetate, b&w ; 8.7 x 12.1 cm.</td>
</tr>
<tr>
<td>http://nla.gov.au/nla.pic-an23417397</td>
</tr>
</table>
</div>

Let’s start to use RDFa to start to build speciﬁc relationships on the page. Let’s start with our subject.

xhtml-rdfa-1.dtd">

<h1 property="dc:title">[Sydney Harbour scene with naval vessels in foreground, and
Harbour Bridge in distance] [picture] /</h1>
<table>
<tr>
<th>Date</th>
<td property="dc:date">[between 1910 and 1962]</td>
</tr>
<tr>
<th>Format</th>
<td property="dc:format">1 negative : acetate, b&w ; 8.7 x 12.1 cm.</td>
</tr>
<tr>
<td property="dc:identifier">http://nla.gov.au/nla.pic-an23417397</td>
</tr>
</table>
</div>

Predicates as attributes

xhtml-rdfa-1.dtd">

<h1 property="dc:title">[Sydney Harbour scene with naval vessels in foreground, and
Harbour Bridge in distance] [picture] /</h1>
<table>
<tr>
<th>Date</th>
<td property="dc:date">[between 1910 and 1962]</td>
</tr>
<tr>
<th>Format</th>
<td property="dc:format">1 negative : acetate, b&w ; 8.7 x 12.1 cm.</td>
</tr>
<tr>
<td property="dc:identifier">http://nla.gov.au/nla.pic-an23417397</td>
</tr>
</table>
</div>

And the objects are the values marked up within the tags.

This is a little geo-explorer application I built. The underlying code has been built incorporating RDFa. As you can see visually, it
looks just like any normal webpage, there is nothing special about it.

By using a Firefox plugin (Fuzz) I can see the underlying semantic structure that exists beyond the visual - what a machine
would see.

RDF
Exterior of.... -37.816068

144.961201
Title
Latitude
Museum
Longitude

Coverage http://nla.gov.au/nla.pic-an25043941

Creator

Sievers, Wolfgang TypeOf

Person
isPrimaryTopicOf

http://nla.gov.au/nla.party-558817

It’s a bit easier to see the semantic relationships as a graph. You’ll often see this sort of representation of semantics. You’ll
notice that there’s now semantic linkages between 2 different URL’s - neat.

OWL
Web Ontology language

Used to describe the meanings of terms in
vocabularies and the relationships between
the terms. Used for applications rather than
human presentation.

OWL - complex rules deﬁning the properties.

What are the capital cities in
countries within Africa?

PREFIX abc: <http://example.com/exampleOntology#>
SELECT ?capital ?country
WHERE {
?x abc:cityname ?capital ;
abc:isCapitalOf ?y .
?y abc:countryname ?country ;
abc:isInContinent abc:Africa .
}

http://en.wikipedia.org/wiki/SPARQL
An example query that returns all the capital cities in countries within Africa. You might want to query all the books that were
published by a company where the authors were born in Melbourne.

Markup HTML using RDFa

Provide RDF output in API’s

What can libraries start to do? This leads to data reuse, mashups, applications. Increased Google rank.

Use external resources such as dbpedia.org - Community effort to extract data from wikipedia & make it available. This is
entering the world of linked data.

See the different spellings of Canberra - we don’t need to know all of these, but can extract them from other resources - linked
data.

For further info, there’s a couple of really good presentations from Sir Tim Berners-Lee on TED

Thanks!
phagon@nla.gov.au
@paulhagon

Thank you.

Introduction to-the-semantic-web

More Related Content

Viewers also liked (9)

More from Hugueins Yánez Rueda (12)

Recently uploaded (20)

Introduction to-the-semantic-web