25.01.2014 Views

Ontology Modeling Profile, an Extension for the ... - ResearchGate

Ontology Modeling Profile, an Extension for the ... - ResearchGate

Ontology Modeling Profile, an Extension for the ... - ResearchGate

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

International Journal of Computer Applications (0975 – 8887)<br />

Volume 6– No.12, September 2010<br />

Two-way mappings between OWL <strong>an</strong>d ODM, ODM <strong>an</strong>d<br />

<strong>Ontology</strong> UML <strong>Profile</strong> <strong>an</strong>d from <strong>Ontology</strong> UML <strong>Profile</strong><br />

to o<strong>the</strong>r UML profiles<br />

The OUP developer firstly defined <strong>the</strong> place of ODM <strong>an</strong>d OUP<br />

in <strong>the</strong> context of <strong>the</strong> MDA four layer architecture <strong>an</strong>d identify <strong>the</strong><br />

main OWL concepts. Then, to support ODM, <strong>the</strong>y defined <strong>the</strong><br />

OUP <strong>an</strong>d describe its details. The proposed UML profile enables<br />

usage of <strong>the</strong> well-known UML notation in ontological<br />

engineering more extensively. The purpose of OUP is to enable<br />

<strong>the</strong> use of <strong>the</strong> st<strong>an</strong>dard UML graphical notation <strong>for</strong> developing<br />

ontologies. You c<strong>an</strong> refer to [7] <strong>for</strong> more in<strong>for</strong>mation.<br />

Due to <strong>the</strong> difference between UML <strong>an</strong>d ontology class, as it is<br />

defined in OWL (owl:Class), OUP defined a stereotyped<br />

\textit{} UML classes to model ontology classes.<br />

And m<strong>an</strong>y subclasses of <strong>the</strong> class OWL:Class , stereotyped<br />

, , ,<br />

, <strong>an</strong>d <strong>for</strong><br />

classes namely Enumeration, Union, Intersection, Complement,<br />

Restriction <strong>an</strong>d AllDifferent respectively. Also, <strong>the</strong>re are m<strong>an</strong>y<br />

differences between UML objects <strong>an</strong>d OWL:Individual, OUP has<br />

stereotyped UML objects as .<br />

Property in OWL is a st<strong>an</strong>dalone concept, different from <strong>the</strong><br />

property concept used in UML. Because of that, OUP used <strong>the</strong><br />

st<strong>an</strong>d alone concept in UML, class, stereotyped by<br />

<strong>for</strong> properties among individuals <strong>an</strong>d<br />

<strong>for</strong> properties among individuals <strong>an</strong>d<br />

primitive data type.<br />

OWL statements are represented via concrete links between<br />

ontology inst<strong>an</strong>ces, individuals. In OUP, Statement is<br />

represented as <strong>an</strong> Object , "ObjectProperty" or<br />

"DatatypeProperty", with two Links - <strong>the</strong> subject Link <strong>an</strong>d <strong>the</strong><br />

object Link. Unlike o<strong>the</strong>r MDA-based approaches to ontology<br />

development both ODM <strong>an</strong>d OUP support modeling of body of<br />

knowledge (i.e. class inst<strong>an</strong>ces) [9].<br />

We have chosen this profile <strong>an</strong>d <strong>the</strong>n updated it because of m<strong>an</strong>y<br />

reasons:<br />

OUP is specialized <strong>for</strong> modeling <strong>the</strong> recent W3C ef<strong>for</strong>t<br />

(OWL).<br />

It has coved almost all <strong>the</strong> practical part of building<br />

ontology,(e.g: Statements).<br />

It shows in details how to use UML modeling tools in<br />

developing <strong>Ontology</strong> using <strong>the</strong>ir profile. For more<br />

details you c<strong>an</strong> refer to [8].<br />

5. THE PROPOSED ONTOLOGY<br />

MODELING PROFILE (OMP)<br />

From <strong>the</strong> above expl<strong>an</strong>ation of <strong>the</strong> OUP, we have noticed that <strong>the</strong><br />

OUP was a complete <strong>an</strong>d worth work not to be neglected. The<br />

ef<strong>for</strong>t spent <strong>the</strong>ir could be accomp<strong>an</strong>ied with our extra <strong>an</strong>d<br />

modified stereotypes to generate a valid <strong>an</strong>d practice oriented<br />

profile.<br />

5.1 OWLDataR<strong>an</strong>ge<br />

Through <strong>the</strong> <strong>an</strong>alysis of <strong>the</strong> ODM, Enumeration are represented<br />

in two m<strong>an</strong>ner. The class called enumeration, which is<br />

stereotyped in <strong>the</strong> OUP. Enumeration is a<br />

class where its elements are individuals. This is due to <strong>the</strong> fact<br />

that in <strong>Ontology</strong>, a class is a set of individuals. The second type<br />

of Enumeration mentioned in <strong>the</strong> ODM was, DataR<strong>an</strong>ge, in<br />

which elements are all primitive literals(e.g Integer, String,<br />

Boole<strong>an</strong> ,.)<br />

Figure 5: OWLDataR<strong>an</strong>ge in <strong>the</strong> OWL metamodel.<br />

5.2 OWLStatement<br />

Statement in OWL, is inherited from RDFStatement. From <strong>the</strong><br />

definition of RDFStatement, we c<strong>an</strong> see that all three<br />

associations, subject, object <strong>an</strong>d predicate, link RDFStatement to<br />

RDFSResource as shown in Figure 6. This design is wide open<br />

to <strong>an</strong>y kinds of statement, even one that does not me<strong>an</strong> <strong>an</strong>ything.<br />

There<strong>for</strong>e, <strong>the</strong> predicate should usually be a resource that<br />

represents a verb, such as "be" or "memorize", or some<br />

characteristic of a resource, like "name". In <strong>the</strong> RDFS MS, such a<br />

resource is <strong>an</strong> rdf:Property, a concept that represents a type of<br />

relationship between resources [17].<br />

Figure 6: OWLStatement in <strong>the</strong> OWL metamodel.<br />

From <strong>the</strong> above definition of a statement in <strong>the</strong> ODM<br />

metamodel, we think that statement in ODM is not supposed to<br />

have a similar concept in UML. What is being represented is its<br />

elements, Subject ,object <strong>an</strong>d Predicate. The OUP has preferred<br />

to use link, <strong>an</strong>d link ends. In our opinion , Subject, object <strong>an</strong>d<br />

predicate are not links, <strong>the</strong>y are individuals <strong>an</strong>d Object property<br />

inst<strong>an</strong>ces. A statement in OWL has to link 3 inst<strong>an</strong>ces, each<br />

plays a role in this statement. If subject is <strong>an</strong> Individual,<br />

predicate could be <strong>an</strong> inst<strong>an</strong>ce of <strong>an</strong> object property , <strong>an</strong>d <strong>the</strong><br />

object could be <strong>an</strong>o<strong>the</strong>r individual or <strong>the</strong> same subject individual<br />

in case of symmetric object properties. So <strong>the</strong> proposed OMP<br />

define stereotyped inst<strong>an</strong>ce specification named <br />

of a class stereotyped <strong>an</strong>d a stereotyped<br />

inst<strong>an</strong>ce specification named <strong>an</strong>d <strong>for</strong> a<br />

class or .<br />

25

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!