CS-566 WEB SEMANTICS

pogonotomygobbleΤεχνίτη Νοημοσύνη και Ρομποτική

15 Νοε 2013 (πριν από 3 χρόνια και 10 μήνες)

140 εμφανίσεις

UNIVERSITY OF CRETE

COMPUTER SCIENCE DEP
ARTMENT





CS
-
566

WEB SEMANTICS






ATHINA TZIAKI (MET)

ANTONIS MISARGOPOULO
S (MET)




MARCH 27
TH
, 2003


Aidministrator Netherlands bv.

Athina Tziaki, Antonis Misargopoulos





2




Abstract


In this
work
, we present
Aidministrator
, a leading
company in the Semantic Web area based in T
he
Netherlands. There is an extend description of
Aidministrator

products and the technology

of
RDF/RDF Schema

is used. In addition, we

compare
RQL with other query languages and finally we
present cases where products are applied.







































Aidministrator Netherlands bv.

Athina Tziaki, Antonis Misargopoulos





3




TABLE OF CONTENTS


1

Introduction

................................
................................
...............................

4

2

Aidministrator Products Overview

................................
.............................

4

2.1

Spectacle Analysis

................................
................................
............

5

2.2

Spectacle e
-
Commerce

................................
................................
.....

7

2.3

Spectacle Enterprise

................................
................................
.........

8

2.4

Sesame: RDF/RDFS storage and retrieval

................................
........

8

2.4.1

Description

................................
................................
.................

8

2.4.2

Sesame’s Architecture Overview

................................
................

9

3.1

RDF

................................
................................
................................
.

11

3.2

RDF Schema

................................
................................
...................

13

3.2.1

Querying at the syntactic level

................................
..................

14

3.2.2

Querying at the structure level

................................
..................

15

3.2.3

Querying at the semantic level
-

The Query Language RQL

....

16

3.3

RQL vs. other Query Languages

................................
.....................

17

3.4

Sesame vs. competitive Products

................................
....................

19

4

Aidministrator’s Portf
olio

................................
................................
.........

20

4.1

Bouwradius Information Monitor

................................
......................

20

4.2

Theater de Flint Website

................................
................................
..

20

3.5

MegaSportZone

Website

................................
................................
.

21

4.3

On
-
To
-
Knowledge Project

................................
...............................

21

4.4

VacanceSelect Website

................................
................................
...

22

5

Conclusion

................................
................................
.......................

23

6

References

................................
................................
.............................

24























Aidministrator Netherlands bv.

Athina Tziaki, Antonis Misargopoulos





4

1

Introduction


Aidministrator, based in The Netherlands, is one of the leading comp
anies in
the process of creating the Semantic Web. The Semantic Web will be more
intelligent than the current web: it will not only enable people to actually find
information, apart from finding it more easily, it will also have intelligence
which it will
use to render useful information to people.


Aidministrator's main activity, product software development, aims at
using the latest technology available to solve real life problems in customer
projects. Of course Aidministrator's participation in technolog
y research
projects is of vital importance to ensure that better solutions will become
available in the near future. Aidministrator's software development is based on
Advanced Java Software Engineering and Artificial Intelligence, and therefore
close ties
with university research and projects sponsored by the European
Union are considered of vital importance to keep up with new developments.




In the following section, there is an extend description of Aidministrator
products, Spectacle and Sesame. In sect
ion 3, we make a brief presentation
of R
esource
Q
uery
L
anguage

and then a comparison with other query
languages.
Finally, in section 4 there are cases where Aidministrator’s
products are applied.


2

Aidministrator Products Overview


Products of Aidministrato
r, like
Spectacle

and
Sesame
, intelligently give
access to and present information within the Semantic Web. Spectacle
organises and structures information and generates websites, intranets and
extranets with the following goal in mind: present information
in such a way
that people find what they need! Spectacle is used in knowledge management
and e
-
commerce projects resulting in successful solutions.



Sesame, the storage and retrieval building block of the Semantic Web
is based on the new semantic web lang
uage RDF, or Resource Description
Language. Sesame is distributed by Aidministrator as an open source product
and therefore available free of charge for anyone who has an interest in
contributing to the creation of the Semantic Web.

A brief description of
Spectacle and Sesame is presented in
Table
.
1
.












Aidministrator Netherlands bv.

Athina Tziaki, Antonis Misargopoulos





5

Product

Description

Customer

Sesame

Semantic Web middleware. Stores
and retrieves RDF (Resource
Description Framework) and RDF
-
Schema annotated information.
Developed as open source software.

Semant
ic

Web develope
rs

Spectacle

E
-
commerce

Semantic navigation for web
-
based
stores. Converts visitors into buyers
through intuitive translation of
customer wishes into product offers.

E
-
commerce companies

Spectacle Enterprise

Thesaurus and taxonomy
-
based
in
formation disclosure. Discloses
enterprise information resources
based on understanding of the context
of the user.

Companies with valuable
information repositories

Spectacle Analysis

Qualitative and quantitative graphical
analysis of information. Helps
i
nformation professionals understand
their information collections.

Knowledge intensive
companies, R&D
departments


Table1.

Aidministrator Products overview





In the following sub
-
section, we present Spectacle
in three different
types

and Sesame
.


2.1

Spectacle Analysis


Spectacle Analysis is a development tool that enables applications for in
-
depth analysis of complex and/or large information repositories.


Visualization provides the opportunity to quickly gain insight in large
volu
mes of information. In a single glance one can easily spot trends and
patterns in the information and detect exceptional cases.



Spectacle Analysis visually shows the cohesion within large collections
of information, typically (but not limited to) collect
ions of documents. It
integrates structured and unstructured information from document
repositories, Web sites, databases, and other applications.


First, one or more characteristics of the domain (also referred to as
classes or types) are defined. Then al
l documents are assigned to one or
more of these types. The Cluster Map
visualizes

which documents belong to
which types. Since documents often belong to more than one type, clusters of
similarly classified documents emerge
as shown in
Fig.

and Fig.2
.




V
isualizing information with Aidministrator Cluster Ma
ps gives you the
opportunity to
see

structure within a domain.
It is useful in situations where
the user has

an analytical task to perform.
Cluster Maps show relations that
are hard to present in text al
one.



Aidministrator Netherlands bv.

Athina Tziaki, Antonis Misargopoulos





6


Fig.1

A Cluster Map
for

"Electrical
-
powered Cars", "Traffic Congestion" and
"Environment"






Fig.2

A Cluster Map, showing documents from an educational domain


Aidministrator Netherlands bv.

Athina Tziaki, Antonis Misargopoulos





7


For clients
,

Cluster Map software provides an animation that allows the
user to vis
ually and conveniently browse through the hierarchy of classes
.
Classes are shown
at the left
, while clustering results are at the right of the
Spectacle Cluster Map Viewer

as shown in Fig.3.




Fig.3

Spectacle Cluster Map Viewer snapshot



2.2

Spectacle e
-
C
ommerce


Spectacle E
-
commerce is a development platform for the creation of website
applications that shows the products of a virtual store in a new way. It helps
visitors to find products they really want by sharing knowledge about the
products in the sto
re. Just as a human shop assistant would do, Spectacle
makes it possible for visitors to explore a virtual store in order to learn what is
there. By displaying products on multiple places the chance for a visitor to find
something increases.


The visitor u
ses Spectacle to learn what is in a virtual store and to
refine the initial whish. Visitors don't have to know exactly what they are
looking for: they look around to learn what is there. The way the products are
Aidministrator Netherlands bv.

Athina Tziaki, Antonis Misargopoulos





8

displayed in the shop refines their ideas or

gives them new ideas about what
to buy.


On the other hand,
shop owner uses Specta
cle as a sales instrument. V
isitor
becomes a customer because the shop shows what is in store. Visitors don't
go to empty shelves, because Spectacle shows alternatives for p
roducts that
are not there. Spectacle helps your potential customer to learn what he or she
wants.
It helps to refine their initial wish.


M
yvacation

Web page
1
is an example of Spectacle E
-
commerce applied to
the domain of travel websites.



2.3

Spectacle Ente
rprise


Spectacle Enterprise is an instrument that
organizations

use to inform their
employees and customers.
R
ight information
is

necessary
to make the right
decisions. Spectacle Enterprise is a unique tool that keeps track of the latest
information by
au
tomatically classifying
new information. It is easy to use and
gives a clear overview of relevant articles and publications.


The information
is
organized in a
comprehensive way.

The user
selects information in a small number of steps and uses a personal
channel to
find interesting information.



Users of websites or intranet generated with Spectacle Enterprise find
information organized in a structure (taxonomy) that is familiar and related to
their tasks. For example, students find new information organi
zed
by courses

they take, but teachers find information organized
by topic
. Spectacle
Enterprise enables users to easily explore or learn what is there.


Organiz
ations benefit from Spectacle Enterprise by
its

outstanding way
of presen
ting information. Thi
s enables workers to
find

and
share

information.
The process of adding new information to the system is simple with the help
of advanced automatic classification software. This software
recognizes

the
topics within a document and classifies them accordingl
y.



2.4

Sesame:

RDF/RDFS storage and retrieval

Sesame is an architecture that allows persistent storage of RDF data and
schema information and subsequent

querying of that information. In section
2.4.1, we present a

description

of Sesame
’s features and

i
n the

section

following that, we look in more detail at several components

of its architecture
.


2.4.1

Description

Sesame is a core component of the
Semantic Web
: it enables agents and
applications to access information intelligently on behalf of the human user.

1
http://demo.aidministrator.nl/myvacation.html

Aidministrator Netherlands bv.

Athina Tziaki, Antonis Misargopoulos





9


T
he next generation of Internet applications, commonly known as the
Semantic Web, requires machine accessible semantics. In order to have
machine support for human tasks, actions, requests, the machine should
"know" the implications. For instance the machin
e should "know" the
implications of "going to the theatre": it might involve hiring a taxi, ordering a
ticket, selection of a performance and so on.


RDF/RDFS is the worldwide standard for expressing machine
accessible semantics. Sesame is one of the first

persistent storages for
RDF/RDFS in the world. Sesame is the first to support the RQL query
language

(see sub
-
section 3.1)
for access of the stored information and its
schema.


So, Sesame is a repository in which machine accessible information is
stored.
Machines are able to perform intelligent ta
sks based on that
information.




There is a second important use of information stored in Sesame:
presentation of available information to human users.


The wealth of stored information in Sesame is a perfect ba
sis for
semantic presentation, navigation and personalization. The combination of
Sesame and Spectacle achieves this. Sesame contains such high
-
level
semantic information that it can be presented to human users in a convenient
way only by a presentation en
vironment as rich and powerful as Spectacle.


Sesame is being developed as Open Source Software by
Aidministrator and the Sesame community, in cooperation with NLNet
2
.

The
foundation
NLN
et stimulates network research and development in the
domain of Intern
et technology
.


2.4.2

Sesame
’s

Architecture

Overview

Sesame’s architecture is shown in Fig.4.

For persistent storage of RDF data,
Sesame needs a
scalable repository
. Naturally, a Data Base Management
System (DBMS) comes to mind, as these have been used for decad
es for
storing large quantities

of data. In these decades, a large number of DBMS's
have been developed, each having their own strengths

and weaknesses,
targeted platforms, and API's. Also, for each of these DBMS's, the RDF data
can be stored

in numerous w
ays.


As we would like to keep Sesame DBMS
-
independent and it is
impossible to know which way of storing

the data is best fi
tted for which
DBMS, all DBMS
-
speci
fi
c code is concentrated in a single architectural layer

of Sesame: the
Repository Abstraction La
yer
RAL (see sub
-
section 4.2 in [2])
.


This RAL is an interface that offers RDF
-
speci
fi
c methods to its clients
and translates these methods to

calls to its specifi
c DBMS.
An important
advantage of the introduction of such a separate layer is that

it makes

it
possible to implement Sesame on top of a wide variety of repositories without
changing any

of Sesame's other components.



In addition, a
n important feature of the RAL is that it is possible to put
one on top of the other. To Sesame's functional

modu
les (the admin, query
and export modules) this is completely transparent, as they will only
communicate

with the RAL at the top of the stack. The RAL at the top can
perform some action when

the modules make calls to it, and then forward
2
http://www.nlnet.nl/

Aidministrator Netherlands bv.

Athina Tziaki, Antonis Misargopoulos





10

these calls to the
RAL beneath it. This process continues

until one of the

RALs
fi
nally handles the actual retrieval request, propa
gating the result back
up again as shown in Fig.5.


The following is a list of possible concrete implementations of the
repository, each with th
eir

own advantages
:



DMBSs
:
Any kind of database can be used: relational databases
(RDBMS), object
-
relational databases (ORDBMS), etc.



Existing RDF stores
:
A number of RDF stores are currently in
development. Sesame can use such a
n RDF store if a RAL is wri
tten
that
knows how to talk to that

speci
fi
c RDF store.



RDF files
:
Files containing RDF can
be used as repositories too. A fl
at
fi
le
is not very practical on its own, as

it will be painfully slow in storing and
retrieving data. However, when combined with
a RAL that

caches all of the
data in memory it becomes a good alternative for small volumes of data.



RDF network services
:
Apart from performance, there is no need for the
repository to be located close to Sesame. Any

network service that offers
basic func
tionality for storing, retrieving and deleting RDF data can be

used by Sesame. An example of a system offering such functionality is, of
course, Sesame itself.

Many of the RDF stores mentioned above can also
be approached as Web services.



Sesame's functi
onal modules are clients of the RAL. Currently, there
are three such modules

(see section 5 in [2])
:



RQL query module
: evaluates RQL queries posed by user



RDF administrator module
: allows incremental uploading and deleting
RDF data and schema information



RDF export module
: allows the extraction of the complete schema and/or
data from a model in RDF format






Fig
.
4. Sesame’s architecture



Fig
.
5. RALs can be stacked to add









functionality

Aidministrator Netherlands bv.

Athina Tziaki, Antonis Misargopoulos





11


Depending on the environment in which it

is deployed, different ways to
communicate with the Sesame

modules may be desirable. For example,
communication over
HTTP

may be preferable in a Web context,

but in other
contexts protocols such as
RMI

(Remote Method Invocation) or
SOAP

(Simple Object Acc
ess

Protocol) may be more suited.




In order to allow maximal
flexibility, the actual handling of these
protocols has been placed
outside the scope of the functional modules
.
Instead, protocol handlers are provided as intermediaries between the

modules an
d their clients, each handling a speci
fi
c protocol.

The introduction
of the repository abstraction layer and the protocol handlers makes Sesame
into

a generic architecture for RDF/S storage and querying, rather than just a
particular implementation

of such

a system.




3


RDF
/

RDF Schema

Overview


RDF is a W3C recommendation that was originally designed to standardize
the definition and use of metadata
-
descriptions of Web
-
based resources.
However, RDF is equally well suited for representing arbitrary data, b
e they
meta
-
data or not.



3.1

RDF


The basic building block in RDF is an object
-
attribute
-
value triple, commonly
written as A(O,V). That is, an object O has an attribute A with value V.
Another way to think of this relationship is as a labeled edge between tw
o
nodes: [O]


A → [V].

This notation is useful because RDF allows objects and values to be
interchanged. Thus, any object from one triple can play the role of a value in
another triple, which amounts to chaining two labeled edges in a graphic
repre
sentati
on. The graph in figure 6

for example, expresses the following
relationships:



hasName

('http://www.csprofessor
s
.org/antwniou/grhgorhs',

"Grhgorhs Antwniou")

teaches

('http://www.csprofessor
s
.org/antwniou/grhgorhs',

http://www.courses.org/CS566')

title

('
http://www.courses.org/CS566',

"Semantic Web")




Aidministrator Netherlands bv.

Athina Tziaki, Antonis Misargopoulos





12




teaches











title








Fig.6.

A
RDF data graph, capturing three statements


RDF also allows a form of reification in which any RDF statement itself can be
the object or value of a

triple. This means graphs can be nested as well as
chained. On the Web this allows us, for example, to express doubt or support
for statements created by other people. Finally, it is possible to indicate that a
given object is of a certain type, such as s
tating that “CS566” is of the type
Course, by creating a type edge referring to the Course definition in an RDF
schema:


type

('http://www.courses.org/CS566',

'http://www.description.org/schema#Course')


The RDF Model and Syntax specification also proposes

an XML syntax for
RDF data models. One possible serialization of the above relations in this
syntax, would look like this:


<rdf:Description

rdf:about="http://www.csprofessor
s
.org/antwniou/grhgorhs">

<s:hasName>Grhgorhs Antwniou</s:hasName>

<s:t
eaches r
df:resource="http://www.
courses.org/CS566"/>

</rdf:Description>


<rdf:Des
cription rdf:about="http://www.
courses.org/CS566">

<s:title>Semantic Web</s:title>

<rdf:type rdf:resource="http://www.description.org/schema#Course"/>

</rdf:Description>



Since the p
roposed XML syntax allows many alternative ways of
writing down information (and indeed still other syntaxes may be introduced),
the above XML syntax is just one of many possibilities of writing down an RDF
model in XML.

It is important to note that RDF is

designed to provide a basic object
-
attribute
-
value model for Web
-
data. Othe
r than this intended semantics,
described o
nly informally in the standard,
RDF makes no data modeling
commitments. In particular, no reserved terms are defined for further data
…/antwniou/grhgorhs

…/CS566

Grhgorhs Antwniou

Semantic Web

hasName

Aidministrator Netherlands bv.

Athina Tziaki, Antonis Misargopoulos





13

mod
eling. As with XML, the RDF data model provides no mechanisms for
declaring vocabulary that is to be used.


3.2

RDF Schema


RDF Schema is a mechanism that lets developers define a particular
vocabulary for RDF data (such as teaches) and specify the kinds of ob
jects to
which these attributes can be applied (such as Professor). RDF Schema does
this by pre
-
specifying some terminology, such as Class, subClassOf and
Property, which can then be used in application
-
specific schemata. RDF
Schema expressions are also va
lid RDF expressions
-

in fact, the only
difference with ‘normal' RDF expressions is that in RDF Schema an
agreement is made on the
semantics
of certain terms and thus on the
interpretation
of certain statements. For example, the subClassOf property
allows
the developer to specify the hierarchical organization of classes.
Objects can be declared to be instances of these classes using the type
property. Constraints on the use of properties can be specified using domain
and range constructs.

Above the dotted
line, we see an example RDF schema that defines
vocabulary for the RDF example we saw earlier:
Course, Professor

and
CSProfessor

are introduced as classes, and
teaches

is introduced as a
property. A specific instance is described in terms of this vocabular
y below the
dotted line.
















Fig.7.

A RDF Schema, defining vocabulary and a class hierarchy



RDF documents and RDF schemata can be considered at three different
levels of abstraction:

1.

at the
syntactic level
th
ey are XML doc
uments

2.

at the
structure level
they consist of a set of triples

…/antwniou/grhgorhs

…/CS566

Professor

Course

CSProfessor

teaches

Type

Type

SubClassO
f

domain

range

Schema

Data

teaches

Aidministrator Netherlands bv.

Athina Tziaki, Antonis Misargopoulos





14

3.

at the
semantic level
they constitute one or more graphs with partially
predefined semantics.


We can query these documents at each of these three levels. We will
briefly consider the pros
and cons of doing so for each level in the next
sections. This will lead us to conclude that RDF(S) documents should really
be queried at the semantic level. We will briefly discuss RQL, a language for
querying RDF(S) documents at the semantic level.



3.2.1

Que
rying at the syntactic level


Any RDF model (and therefore any RDF schema) can be written down in XML
notation. It would therefore seem reasonable to assume that we can query
RDF using an XML query language. However, this approach disregards the
fact that
RDF is not just an XML dialect, but has its own data model that is
very different from the XML tree structure. Relationships in the RDF data
model that are not apparent from the XML tree structure become very hard to
query.

As an example, let us look again

at the XML description of the RDF model:


<rdf:Description

rdf:about="http://www.csprofessor
s
.org/antwniou/grhgorhs">

<s:hasName>Grhgorhs Antwniou</s:hasName>

<s:te
aches rdf:resource="http://www.
courses.org/CS566"/>

</rdf:Description>


<rdf:De
scription
rdf:about="http://www.
courses.org/CS566">

<s:title>Semantic Web</s:title>

<rdf:type rdf:resource="http://www.description.org/schema#Course"/>

</rdf:Description>


In an XML query language such as XQuery [Chamberlin et al., 2001],
expressions to traverse the

data structure are tailored towards traversing a
node
-
labeled tree. However, the RDF data model in this instance is a graph,
not a tree, and moreover, both its edges (properties) and its nodes
(subjects/objects) are labeled. In querying at the syntax leve
l, this is literally
left as an exercise for the query builder: one cannot query the relation
between the resource signifying ‘Grhgorhs Antwniou’ and the resource
signifying ‘Semantic Web’ without knowledge of the syntax that was used to
encode the RDF dat
a in XML.

Ideally, we would want to formulate a query like “Give me all the
relationships that exist between Grhgorhs Antwniou and The Semantic Web”.
However, using only the XML syntax, we are stuck with formulating an
awkward query like “Give me all the e
lements nested in a Description element
wit
h an about attribute with value
'http://www.csprofessor
s
.org/antwniou/

grhgorhs’, of which the value of its resource attribute occurs elsewhere as the
about attribute value of a Description element which has a nes
ted element
title with the value ‘Semantic Web’.”

Aidministrator Netherlands bv.

Athina Tziaki, Antonis Misargopoulos





15

Not only is this approach inconvenient, it also disregards the fact that
the XML syntax for RDF is not unique: different ways of encoding the same
information in XML are possible and in use currently. This
means that one
query will never be guaranteed to retrieve all the answers from an RDF
model.



3.2.2

Querying at the structure level


When we abstract from the (XML linearization, or any other) syntax, any RDF
document represents a set of triples, each triple re
presenting a statement of
the form Object
-
Attribute
-
Value. A number of query languages have been
proposed and implemented that regard RDF documents as such a se
t of
triples, and that allow
querying

such a triple set in various ways.

Our RDF/RDF Schema exa
mple corresponds to the following set of triples:


(type Course Class)

(type Professor Class)

(type CSProfessor Class)

(subClassOf CSProfessor Professor)

(type teaches Property)

(domain teaches Professor)

(range teaches Course)

(type antwniou/grhgorhs CSPr
ofessor)

(type CS566 Course)

(teaches antwniou/grhgorhs CS566)


An RDF query language such as, for example, Squish [Miller, 2001]
would allow us to query which resources are known to be of type
CSProfessor:


SELECT

?x

FROM

somesource

WHERE

(r
df::type
?x C
SProfessor)



The clear advantage of such a query is that it directly addresses the
RDF data model, and that it is therefore independent of the specific XML
syntax that has been chosen to represent the data.


However, a shortcoming of Squish or any query l
anguage at this level
is that it interprets
any
RDF model only as a set of triples, including those
elements which have been given a special semantics in RDF Schema. For
example, since http://www.csprofessors.org/antwniou/grhgorhs is of type
CSProfessor, a
nd since CSProfessor is a subclass of Professor,
http://www.csprofessors.org/antwniou/grhgorhs is also of type Professor, by
virtue of the intended RDF Schema semantics of type and subClassOf.


However, there is no triple that explicitly asserts this fact
. As a result, the
query


Aidministrator Netherlands bv.

Athina Tziaki, Antonis Misargopoulos





16

SELECT

?x

FROM

somesource

WHERE

(rdf::type ?x Professor)


will fail because the query only looks for explicit triples in the store, whereas
the triple (type antwniou/grhgorhs Professor) is not necessarily present in the
store, bu
t is implied by the semantics of RDF Schema. Notice that simply
expanding the query into something like



SELECT

?x

FROM


somesource

WHERE

(rdf::type ?x ?c1),

(rdfs::subClassOf ?c1 Professor)

OR


?c1 ~ Professor


will solve the problem in this specific e
xample, but does not solve the problem
in general.



3.2.3

Querying at the semantic level
-

the

Query Language RQL



RQL is a query language for formulating queries on RDF Schema and Data. It
is being developed within the European IST project C
-
Web and its proje
ct
MESMUSES by the Institute of Computer Science at FORTH in Greece.


RQL is a functional language that adopts the syntax of OQL. The
output of RDF Schema queries is again legal RDF code, which allows the
output of queries to function as input for subsequ
ent queries.

It is defined by a
set of basis queries and iterators, which can be used to build new ones
through functional decomposition.


The core queries are the basic building blocks of RQL, which give
access to the RDF Schema specific contents of an R
DF triple store, with
queries such as Class (retrieving all classes), Property (retrieving all
properties) or Professor (returning all instances of the class with name
Professor). This query returns also all instances of subclasses of Professor,
since thes
e are also instances of the class Professor, by virtue of the
semantics of RDF Schema. We can ask for all direct instances of Professor,
ignoring all instances of subclasses, through the query ^Professor.


RQL can also query the structure of the subclass h
ierarchy. The query
subClassOf(Professor) would return the class CSProfessor as the only result.
In general, this returns all the direct and indirect subclasses of Professor since
RQL is aware of the transitivity of the subclass relation. The query
subClas
sOf^(Professor) would return only the immediate subclass. RQL,
being based on OQL, also allows a select


from


where construct.


RQL relies on a formal graph model that enables the interpretation of
superimposed resource descriptions by representing prop
erties as self
-
existent individuals and introducing a graph instantiation mechanism that
permits multiply classification of resources.


Aidministrator Netherlands bv.

Athina Tziaki, Antonis Misargopoulos





17

For example, the query

select Y

from CSProfessor{X}.teaches{Y}


returns all courses are being taught by CSProfessors e
ffectively doing
pattern
-
matching along a path in the graph of figure 7. Note that RQL path
expressions enable free mixing of data and schema information.



3.3

RQL vs. other Query Languages


Table.2

presents a general comparison for some languages on some ba
sic
criteria describing the language characteristics.



Table
.
2


As it is presented,

most of the query languages rely

on a triple data model

(they use collections of statements of the form <subject, predicate, object> to
encode RDF/S description bases an
d Schemas).
N
evertheless
, RQL follows a
graph data model, which enables

it

to perform more complex queries over the
resource description graphs.

RQL is the only one language that having both the properties of
closure

and
orthogonality
,

supports functional

composition of queries

(results of a
query could be used as input of another query). It also permits any kind of
data as input and output of queries.

RQL
,

however
,

lacks the property of
generality
which is
the ability to
support all the primitives

of the
ontology/metadata model
.

Finally,
RQL
exploits most of

the RDF/S modeling constructs such as
single/multiple
inheritance, single/multiple instantiation, container values and typed literal
values but it does not support reification.

Table.3
hosts a comparis
on of RDF/S query languages according to
five axes:
Modeling constructs
supported,
Ontology Querying
,
Data Querying
,
Aidministrator Netherlands bv.

Athina Tziaki, Antonis Misargopoulos





18

Data/Ontology Querying
and
Additional Features
provided. Regarding the
Modeling constructs
,
m
ost query languages support the basic data typ
es, i.e.
strings and integers, with the exception of RQL, VERSA and RDFQL that
provide more elabora
ted data types (reals, dates, URL
s or enumerated types)

In term of
Ontology Querying

the ability to traverse the
ancestor/descendant hierarchies of classes a
nd properties requires the use of
ontology (schema) knowledge and only RQL and TRIPLE can perform it
transitively. Only RQL can also pose sophisticated filtering conditions on
class/property hierarchies. Apart from (in)equality and “like” conditions on the

names of ontology constructs, RQL can also pose selection filter
s based on
subsumption testing
on the subsumption relation among classes or properties,
and on the namespaces in which these
ontology constructs are defined
.



Table.3


Regarding
the

Data Q
uerying
,
only RQL and VERSA suppo
rt
container
values constructor
s.
Complete Boolean filters such as conjunction, disjunction
and negation are only supported by RQL, RDFQL, VERSA and TRIPLE.


In term of
Data/Ontology Querying
,

only RQL is capable of
incorpo
rating knowledge of ontology. RQL features generalized path
expressions with variables on labels of both nodes and edges. The ability to
perform nested queries is only supported by RQL and VERSA.

RQL and
TRIBLE are
also
the only query languages supporting
existential and
universal quantifiers.


Regarding the
Additional Features

RQL features min, max, count
,
average
and sum functions, as known from relational databases.




Aidministrator Netherlands bv.

Athina Tziaki, Antonis Misargopoulos





19

3.4

Sesame vs. competitive Products



In Table.
4

there is an
evaluation of the technical
features of practical
implementations supporting storage and querying facilities fo
r description
bases and schemas.



Table.4

Sesame vs. competitive Products



As one can see, most of the tools provide a web
-
based demonstration
of the facilities they supp
ort, while open source distribution in the form of
public licenses and the existence of documentation assist third
-
party
developers when building applications on top of these tools. On the contrary,
the lack of tutorials does not facilitate the overview of

the systems.


M
ost of the tools presented can be used with any software platform
(supporting though the Java runtime environment). As we can see from
column
Implementation Language
, 10 out of 14 tools were developed with
Java, thus profiting from the plat
form independence endorsed by Java. The
next most popular implementation language of ontology tool developers is C,
a language whose stability and functionality has already been tested for years
in the deployment of large
-
scale applications.


The populari
ty of Java as an implementation medium can also be
noted when examining the
API support
provided by the tools. However, apart
from Java and C, the developers provide a set of APIs written in different
languages, e.g., Perl, Python and Tcl in an attempt to
facilitate the
Aidministrator Netherlands bv.

Athina Tziaki, Antonis Misargopoulos





20

collaboration of their tool with other applications. These APIs provide
functions for querying and updating description bases and can be used for
interfacing with clients. Hence, they offer third party
-
developers the ability to
deploy their
applications on top of these tools. For this reason, the criterion of
API support can be thought of as a measure of the extensibility and the
degree of collaboration with other applications characterizing by the tool.




4

Aidministrator’s
Portfolio

In this

section, we present some cases of Aidministrator Products
applications.


4.1

Bouwradius
Information Monitor

Bouwradius
3

is a knowledge centre in the Building and Construction sector in
the Netherlands
. A
idministrator has been asked to develop a knowledge
infr
astructure as part of the knowledge management strategy of Bouwradius.
The goal is
to distribute information in a convenient way
. Access to and the
maintenance of information are the current bottlenecks. Valuable information
remains hidden from others in
-

and outside the organization. Bouwradius
wishes to distribute information to different target groups like teachers,
internal employees and coordinators.


The knowledge infrastructure was incrementally built by Aidministrator.
After a few initial workshop
s, the first version of the infrastructure was realized
and evaluated by user panels at Bouwradius. Each target groups has its own
Exploration Context, combining content, navigation and design. Together with
Bouwradius
A
idministrator created different cont
exts that suited the ta
sks of
different target groups.


In order to create an easy input process, automatic classification
software of Irion is used to add new information to the infrastructure. This
software automatically identifies what keywords are rele
vant for new
information based on a large set of training documents.


The knowledge infrastructure has been developed with Spectacle.
Spectacle supports Exploration Contexts based on meta
-
information. Each
profile gets a different view on the information.
The benefits for Bouwradius
are:



user convenience: short time needed to find information, excellent
overview



easy maintenance: Exploration Contexts can be adapted to changing
organization, automatic classification of new information



incremental developm
ent: fast delivery, controllable process

4.2

Theater
de Flint

Website

The website of Theatre de Flint
4
, in Amersfoort
-

the Netherlands, has been
designed and realized by Aidministrator Nederland bv. The site is Spectacle
-
based and kept up
-
to
-
date by the Flint

personnel themselves.



3
http://www.bouwradius.nl/

4
http://www.theaterdeflint.nl/



Aidministrator Netherlands bv.

Athina Tziaki, Antonis Misargopoulos





21

4.3

MegaSportZone

Website

The Intersport MegaStores
5
, located in Alkmaar and Groningen, are among
the largest dedicated sports and outdoor activities retail stores in Europe.
Both stores specialize in providing expert assistance to t
he casual and
professional sportsperson. Additionally, the stores emphasize the fun in sports
by organizing special sport
-
related events and

activities at both locations.


Aidministrator designed, developed and hosts the MegaSportZone on
the Spectacle plat
form, which results in an easy to maintain website, intended
to draw the sport
-

and fun
-
loving public to one of the physical MegaStores.
The site provides information on the enormous amount of sporting and
outdoor clothing and equipment that is available a
t the stores, and informs the
visitors about the special services offered and the special events that are
organized.



4.4

On
-
To
-
Knowledge
Project

On
-
To
-
Knowledge
6

is a project in the Information Society Technologies (IST)
Program for Research, Technology Deve
lopment & Demonstration under the
5th Framework Program. The pr
ogram runs from 1999 to 2002.


Efficient knowledge management has been identified as key to
maintaining the competiveness of organizations. Ontologies have been
developed in the knowledge engi
neering discipline as a means to share and
reuse knowledge. The On
-
To
-
Knowledge project develops methods and tools
and employs the full power of the ontological approach to facilitate knowledge
management. The On
-
To
-
Knowledge tool helps knowledge workers w
ho are
not IT specialists to access company
-
wide information repositories in an
efficient, natural and intuitive way.




One of the results of the project is the
O
ntology
I
nference
L
ayer OIL.

Ontologies provide a shared and common understanding of a domain

that can
be communicated across people and application systems. Therefore, they will
play a major role in supporting information exchange processes in various
areas. However, a prerequisite for such a role is the development of a joint
standard for specif
ying and exchanging ontologies. The Ontology Inference
Layer OIL is a proposal for such a standard.


The
goal

of the On
-
To
-
Knowledge project is to support efficient and
effective knowledge management.
It is

focus
ed

on acquiring, maintaining, and
accessin
g weakly
-
structu
red online information sources as shown in Fig.8

5
http://www.megasportzone.nl/

6
http://www.ontoknowledge.org/






Aidministrator Netherlands bv.

Athina Tziaki, Antonis Misargopoulos





22


Fig.8 On
-
To
-
Knowledge project goal



Acquiring
: Text mining and extraction techniques are applied to
extr
act semantic information from textual information (i.e., to acquire
information).



Maintaining
: RDF and XML are used for describing syntax and
semantics of semi
-
structured information sources. Toolsupport
enables automatic maintenance and view definitions
on this
knowledge.



Accessing
: Pushservices and agent technology support users in
accessing the information.


4.5

VacanceSelect

Website

VacanceSelect
7
, one of the most
successful

and fastest growing Dutch travel
agencies on the Internet, has selected Spectacl
e to present its products to its
customers. The first step in this process is the implementation of the site of
VacanceSelect's bigge
st affiliate, Ilse Media

and is able

to improve on the
conversion rate of their current system, i.e. the percentage of visi
tors
that
actually makes a purchase.


Another reason for selecting Spectacle is pure speed: production tests
have shown that Spectacle is up to 50 times faster than competitor systems.
Spectacle offers visitors the speed they require, and it offers Vacance
Select
the platform for growth.







7
http://vakantie.ilse.nl/




Aidministrator Netherlands bv.

Athina Tziaki, Antonis Misargopoulos





23

5

Conclusion

As a conclusion, we present more information about the location and the
contact links with Aidministrator Netherlands bv.






Address


Aidministrator Netherlands bv.


Prinses julianaplein 14
-
b


3817 cs Ame
rsfoort


The Ne
t
herlands




Phone


+31
-
(0)33
-
4659987(phone)

+31
-
(0)33
-
46 59 987 (fax)




E
-
mail

info@aidministrator.nl


job@aidministrator.nl

(job opportunities)

webmaster@aidministrator.nl

(suggestions or website problems)
































Aidministrator Netherlands bv.

Athina Tziaki, Antonis Misargopoulos





24

6

References




Technical Reports & Papers


[1]

Aimilia Magkanaraki, Grigoris Karvounarakis, Ta Tuan Anh, Vassilis
Chr
istophides, Dimitris Plexousakis,
Ontology Storage and Querying
,
Technical Report 308, ICS
-
FORTH, Heraklion, Crete, Greece, April 2002

[2]

Jeen Broekstra, Arjohn Kampman,
A generic Architecture for Storing
and Querying RDF and RDF schema
, Copyright © 2001
Aidministrator
Netherlands bv.

[3]

Antoniou Grigoris,
A Semantic Web Primer





URLs



www.aidministrator.nl


www.sesame.aidministrator.
nl


www.w3.org/RDF/





























Aidministrator Netherlands bv.

Athina Tziaki, Antonis Misargopoulos





25