21
Knowledge Representation Technologies in the Semantic Web Stephanie Stroka [email protected] Information Technology and Systems Management, Salzburg University of Applied Sciences Abstract. This paper covers technologies for representing knowledge in the Semantic Web. First, the Semantic Web vision, its criticism and its major benefits are introduced. Subsequently, ontologies are defined and presented, structured by their semantically richness. The eXten- sible Markup Language (XML), XML Schema (XMLS), the Resource Description Framework (RDF), RDF Schema (RDFS) and the Web On- tology Language (OWL), which are well-established markup languages and practical technologies used to build ontologies in the Semantic Web, are demonstrated in the following chapter. To complete the logical layer applied in the Semantic Web, queries and rules are explained afterwards. 1 Introduction To understand the principle of the Semantic Web and its Knowledge Represen- tation Technologies it is important to know about the concept of human com- munication. In a conversation a speaker sends syntax, which is a combination of ordered symbols like words and punctuation marks, to one or more auditors to enunciate a denotation of a sentence. The auditors receive the symbols and try to understand the meaning of them. The interpretation is supported by several environment factors, e.g. the previous conversation, the location, the gesture and facial expression of the speaker. In communication theory the denotation of the symbols is called semantics and the interpretation is called pragmatics. The context covers the specific communication topic and how it is related to other terms. Teaching a computer to understand data is a very complex topic of Artificial Intelligence. To enable those conversation processes for machines we have to adapt the concepts of communication theory by defining a syntax that is readable by every computer, making knowledge about objects of interest accessible, and developing software that is able to construe and to use the defined knowledge.

Knowledge Representation Technologies in the Semantic Web

Embed Size (px)

Citation preview

Page 1: Knowledge Representation Technologies in the Semantic Web

Knowledge Representation Technologies in theSemantic Web

Stephanie [email protected]

Information Technology and Systems Management,Salzburg University of Applied Sciences

Abstract. This paper covers technologies for representing knowledgein the Semantic Web. First, the Semantic Web vision, its criticism andits major benefits are introduced. Subsequently, ontologies are definedand presented, structured by their semantically richness. The eXten-sible Markup Language (XML), XML Schema (XMLS), the ResourceDescription Framework (RDF), RDF Schema (RDFS) and the Web On-tology Language (OWL), which are well-established markup languagesand practical technologies used to build ontologies in the Semantic Web,are demonstrated in the following chapter. To complete the logical layerapplied in the Semantic Web, queries and rules are explained afterwards.

1 Introduction

To understand the principle of the Semantic Web and its Knowledge Represen-tation Technologies it is important to know about the concept of human com-munication. In a conversation a speaker sends syntax, which is a combination ofordered symbols like words and punctuation marks, to one or more auditors toenunciate a denotation of a sentence. The auditors receive the symbols and tryto understand the meaning of them. The interpretation is supported by severalenvironment factors, e.g. the previous conversation, the location, the gestureand facial expression of the speaker. In communication theory the denotation ofthe symbols is called semantics and the interpretation is called pragmatics. Thecontext covers the specific communication topic and how it is related to otherterms.

Teaching a computer to understand data is a very complex topic of ArtificialIntelligence. To enable those conversation processes for machines we have toadapt the concepts of communication theory by defining a syntax that is readableby every computer, making knowledge about objects of interest accessible, anddeveloping software that is able to construe and to use the defined knowledge.

Page 2: Knowledge Representation Technologies in the Semantic Web

2

2 The Semantic Web

2.1 A vision of the Web 3.0

The Semantic Web, also known as the Web 3.0, has been described by TimBerners-Lee 1 in [1] as the Web in which computers are capable of analyzing allthe data with its content, links and transactions between people and computers.Tim Berners-Lee envisioned machines that communicate among each other andagents with intelligent behaviour.

Simplified, the adoption of the Semantic Web will enable us to store knowl-edge about web content in a structured form. An illustrative example is givenbelow.

On a server of a university a document could contain information about howthe university hierarchy is organized, e.g. classified in university, rector, depart-ment, head of department, lecturer and student. An occurrence of this schemecan be Salzburg University of Applied Sciences, Dr. Erhard Busek, InformationTechnology & Systems Management, FH-Prof. DI Dr. Thomas Heistracher, DI.Dietmar Glachs and Stephanie Stroka. Furthermore, the classifications are re-lated among each other. A lecturer, for example, teaches one or more students.The predicate teaches forms a relationship between a subject lecturer and anobject student. It is common to say that a predicate is a property of the sub-ject. Those relations are also observed in the document. With defining inferencerules on a property it is also possible to build new connections, e.g. if a studentStephanie Stroka studies Information Technology & Systems Management andFH-Prof. DI Dr. Thomas Heistracher, who works at Salzburg University of Ap-plied Sciences, is the head of the department Information Technology & SystemsManagement, we can infer that Stephanie Stroka studies at Salzburg University ofApplied Sciences. Since the subject, the predicate and the object in the SemanticWeb adoption are defined as Uniform Resource Identifiers (URI) it is possible tocompare these URIs with URIs from other Semantic Web documents, in order toinfer that these entities are equal. Hence, it would be possible to reason that theperson Stephanie Stroka, who has a web site http://www.steffi.com, is actuallyequal to the student Stephanie Stroka, who studies Information Technology &Systems Management at Salzburg University of Applied Sciences.

Moreover, so called software agents that are autonomous applications thatreveal and modify a certain environment (e.g. the World Wide Web) can collectthe semantic data, reason about it, build a new spatial model of the world, andare, therefore, able to conclude transactions or to display demanded content.

The Web 3.0 is an extension to the existing Web and its semantics are encodedinto web pages which make them transparent in normal use. On the one hand thisis an advantage for the ordinary user, because this does not require learning anew acquaintance with the web. On the other hand it can also be a disadvantageas we will see in the next chapter.

1 Tim Berners-Lee is the director of the World Wide Web Consortium (W3C) and thefounder of the World Wide Web

Page 3: Knowledge Representation Technologies in the Semantic Web

3

2.2 Sceptical Reactions

An important factor on why the Semantic Web is still not widespread is, thata lot of companies do not see the benefit of it [2]. They consider that a distinctkeyword declaration coded into their HyperText Markup Language (HTML)sites is adequate. On the one hand, they do not care about web users who cometo their web page searching something completely different, on the other handthey do not know that other users, who would be potential customers, cannotfind their website.

They also hold that the Return of Investment (ROI) is not efficient enough,hence, they are afraid of spending money for a technology that is not yet in useand that gain could not yet been proved.

For companies that have to share their knowledge over distributed branchesanother big criterion is the privacy, because without certain precautions knowl-edge data cannot be prevented from being read by software agents or humans.

The Semantic Web is build on the eXtensible Markup Language (XML), theResource Description Framework (RDF), ontologies and web rules and we willget acquainted with these terms in this paper. Dealing with the issue securityit is obvious that, in addition to security in the layers below the applicationlayer, we need to use XML security, RDF security, secure ontologies and secureWeb Rules to ensure privacy in the Semantic Web. These types of security arepresented in [3].

To overcome that doubts innovative companies have to adopt this new tech-nology to assert that the scepticism is arbitrary. In the next section we willbecome acquainted with some reasons for the Semantic Web.

2.3 Reasons for the Semantic Web

The more information we have on the Internet the more difficult it is to findterms of specific contexts. Simply searching for keywords is not enough anymoredue to incorrect and incomplete keyword declarations and ambiguous words thatexist in our natural languages. The signification of the Semantic Web is to bringstructure into the information chaos and to stem the information overload sothat it becomes possible to find terms of specific contexts.

Furthermore the so called stovepipe systems, which are legacy systems pro-duced and developed to solve a specific problem, can constitute problems indistributed environments. To modify and maintain such hard coded communi-cation is often time-consuming, expensive and sometimes not possible. With theuse of knowledge representing languages adapting, changing and maintainingsystems becomes feasible.

The potential implementation of the Semantic Web is a distinguished on-lineplatform for knowledge management in organisations. With the introduction itbecomes possible to create, personalize, represent and distribute knowledge overbusiness branches with respect to their individual needs.

Page 4: Knowledge Representation Technologies in the Semantic Web

4

2.4 Semantic Web Technologies in use

In Fig.1 we see how the Semantic Web is layered. On the bottom there is the termidentification, which is commonly represented by a URI or an InternationalizedResource Identifier (IRI).

Thereon, a syntax and a structure have to be defined. XML, XMLS and RDFare commonly used technologies.

Ontologies are the core concept of Knowledge Representation in the SemanticWeb. RDFS and OWL are languages that can produce such models.

The logic layer, which enables software agents to reason about the representeddata, is composed through ontologies, queries and rules.

Security mechanisms should be used across every technological layer.

Fig. 1: An abstract version of the Semantic Web Layer Cake [4]

To create a knowledge representing on-line presence we have to build a modelof the spatial world similar to what we have seen in the university structureexample. Simplified, we have to build relations between terms or classes. Thisconcept of building a feasible model is explained in the following chapter.

3 Ontologies

The general idea behind ontologies is to make knowledge explicit by expressingconcepts and their relationships.

Page 5: Knowledge Representation Technologies in the Semantic Web

5

In other words ontologies define the common words and concepts used to de-scribe and represent an area of knowledge or collection of information about dataand how the data is related. Thus, ontologies provide the means for establishinga semantic structure.

Referring to the theory of communication, ontologies would represent thecontext of the terms.

In the Semantic Web ontologies are semi-structured and represent an openworld, which means that the model can grow with the data and that an ontologydoes not contain every existing real world entity. An ontology model can bemerged with another ontology model. Thus, ontologies in the Semantic Web arepartial and modular.

When we talk about ontologies we distinguish between the semantically rich-ness of the various types. In Fig. 2 we can see the common kinds of ontologies,starting at the left bottom where the semantics are simple and weak, going onto the top right where the semantics get more complex and data get logicallyinferable.

Fig. 2: The Ontology Spectrum, adapted from [2]

The subsequent chapters give an introduction to specific kinds of ontologies.

3.1 Taxonomies

A taxonomy is a method to classify or categorize a set of terms in a hierarchicalstructure. In general, it is the study of the general principles of scientific classifi-cation. When we adapt this definition to the domain of information technologywe can say that a taxonomy is the classification of information entities in the

Page 6: Knowledge Representation Technologies in the Semantic Web

6

form of a hierarchy, according to the presumed relationship of the real-worldobjects that they represent.

A taxonomy, in general, is semantically weak, because it does not express richmeaning and does not distinguish between aggregation and generalization/spe-cialization relations. An example of a partial taxonomy is given in Fig. 3.

Fig. 3: A taxonomy with aggregation and generalization/specialization relations

3.2 Thesauri

A thesaurus defines relationships between words and phrases structured in ataxonomy. Some examples of term relations would be synonyms, homonyms, theis narrower than and the is broader than relations.

Homonyms describe two or more equal words with different meanings, whereassynonyms are different words with the same meaning. The relations is nar-rower than and is broader than are relations between a parent and a child sub-classification, where is narrower than declares that the subject is more specificthan the object and vice versa for is broader than (see Fig. 4).

3.3 Conceptual Models

Conceptual Models are common in modeling databases or applications. TheUnified Modeling Language (UML) [5] is a widespread Conceptual Model insoftware engineering. Fig. 5 shows an example of a UML diagram.

Page 7: Knowledge Representation Technologies in the Semantic Web

7

Fig. 4: An example of thesaurus relations

organisationpackage Data[ ]

Network Administrator

-networkKnowledge

Organsiation

-name-commercial register entry-stakeholder-employees

+sell()

-programmingKnowledge

Developer

-managementKnowledge

Manager

Person

-name-age

+beBorn()+passAway()

Employee

+worksAt()

corporation

-shareholder

+buyAShare()

Fig. 5: An example of a class-diagram in UML

Page 8: Knowledge Representation Technologies in the Semantic Web

8

3.4 Logic Theories

Regarding to [6], the function of logic is to define the truth of each logicalsentence in each possible world or model. These collections of assertions arecalled theories [7].

Logic Theories are built on axioms or statements defined in a knowledgebase and inference rules, which together are used to prove theorems about thedomain. With these evidence it is possible to create new knowledge.

A practical benefit of using Logic Theories is that not every relation in anontology has to be defined. In other words, with specified rules and a knowledgebase it is possible to infer new relations. An example was already given in Sec.2.1, where it was inferred that a student Stephanie Stroka studies at SalzburgUniversity of Applied Sciences when she studies Information Technologies &Systems Management, which has FH-Prof. DI Dr. Thomas Heistracher as headof department and when FH-Prof. DI Dr. Thomas Heistracher works at SalzburgUniversity of Applied Sciences.

Another advantage is that we do not have to construct one single, complete,closed world ontology, but we have many spatial ontologies that can be mergedby inferring if ontology components are equal. This is a basic requirement forontologies spread over a big network as the World Wide Web.

Due to the fact that logic theories state about the accuracy of axioms, soft-ware agents can process transactions based on predefined requests of the trans-action purchasers. For instance, in a vacation booking situation the purchaserhas specific desires about the destination, the vacation time, the flight, the hoteland so on. An application based on logic theories can infer if a vacation offerdoes or does not fit to these desires.

In the Semantic Web Logic Theories can be constructed by software agentsthat obtain knowledge from meta data stored in knowledge representing webdocuments.

4 Knowledge Representation Technologies

4.1 What is Knowledge Representation?

Knowledge Representation is a branch of Artificial Intelligence that providesaccess to a structured collection of information and a set of inference rules. Theinformation and rules can then be used for automated reasoning, e.g. with thehelp of software agents.

In this section we will have an introduction to the markup languages used torepresent knowledge.

4.2 Knowledge Representation Languages

XML and XMLS

Page 9: Knowledge Representation Technologies in the Semantic Web

9

As appointed in the introduction of this paper we need a syntax to exchange mes-sages in a communication situation. The eXtensible Markup Language (XML)and the XML validation document XML Schema (XMLS) are used to add arbi-trary structure to the documents. XML is a well-formed markup language whichlets everyone create own tags that may surround a portion of content, but saysnothing about what the structure means.

On the Web XML is often used to store meta data, because it is applicationindependent and has a human readable form, but exchanging XML documentsbetween systems is only reasonable when both systems know what the tagsdenote.

XML is not a knowledge representation language, but the syntax is usedin many knowledge representation languages, e.g. RDF and OWL. Hence, it isimportant to know how an XML file is built.

XMLS is a template and validation document that defines the valid elementsand attributes of an XML file.

For those who are not familiar with the XML and XMLS Syntax please havea look at Listing 3 and 4 (Appendix A).

RDF and RDFS

The Resource Description Framework (RDF) is a W3C Recommendation sinceFebruary 2004 which jointly replaced RDF Model and Syntax (1999 Recom-mendation) and RDF Schema (2000 Candidate Recommendation). It has beendeveloped by the RDF Core Working Group as part of the W3C Semantic WebActivity [8].

RDF was developed with the motivation to provide web meta data and openinformation models, to get new information by combining data from severalapplications and to enable automated processing of web information by softwareagents.

RDF is the foundation layer of the Semantic Web. The semantics are encodedin sets of triples, where each triple consists of a subject, a predicate or propertyand an object, similar to what we have in natural language sentences.

Fig. 2 shows an RDF graph, which is a set of RDF triples. A node may be aURI reference or blank, which means that it is a unique node with no separateform of identification. The object node can also be a literal. The property is alsoa URI.

The graph can be interpreted as follows:

– A subject Henry works at an organisation company x,– Henry is 23 years old and– Henry has some friend who is also 23 years old. We could also say that there

exists a friend of Henry who is also 23 years old.

To store information represented in an RDF graph on-line we translate itinto an RDF/XML file. In Listing 1 we see an example on how the representingRDF/XML file could look like. [9]

Page 10: Knowledge Representation Technologies in the Semantic Web

10

Fig. 6: RDF triple relations between resources, literals and blank-nodes

Listing 1: RDF/XML example

<?xml version=” 1 .0 ”?><rdf:RDF xmlns : rd f=” ht tp : //www.w3 . org /1999/02/22− rdf−

syntax−ns#”xmlns:ab=” ht tp : //www. about . com/”xml:base=” ht tp : //www. henrys page . com/”>

<r d f :D e s c r i p t i o n rd f : ID=”Henry”ab:work=” ht tp : //www. job . com/”ab:age=”23”><ab : f r i e nd rdf :nodeID=” s3 f o ” />

</ r d f :D e s c r i p t i o n>

<r d f :D e s c r i p t i o n rdf :nodeID=” s3 f o ”ab:age=”23”>

</ r d f :D e s c r i p t i o n></rdf:RDF>

The Resource Description Language Schema (RDFS) is a vocabulary lan-guage that provides the users to define terms they intend to use in their RDFdocument[10], similar to the design in object oriented programming (OOP) lan-guages. It differs from OOP in that it describes properties in terms of theirclasses of resource to which they apply. The four most important RDFS vocabu-

Page 11: Knowledge Representation Technologies in the Semantic Web

11

lary definitions are rdfs:Class, rdf:Property, rdfs:domain and rdfs:range.They define which nodes are connected through a certain property.

In Listing 4 we defined an RDFS for the Henry works at company x triple inour previous RDF example.

Listing 2: RDFS example

<?xml version=” 1 .0 ”?><rdf:RDF

xmlns : rd f=” ht tp : //www.w3 . org /1999/02/22− rdf−syntax−ns#”xmlns : rd f s=” ht tp : //www.w3 . org /2000/01/ rdf−schema#”>

<r d f s : C l a s s rd f : about=” ht tp : //www. example . com/employee”>

< r d f s : l a b e l>Employee</ r d f s : l a b e l><rdfs:comment>A Person who works somewhere</

rdfs:comment><r f d s : s u b c l a s sO f r d f : r e s o u r c e=” ht tp : //www. example . com

/person ”/></ r d f s : C l a s s>

<r d f s : C l a s s rd f : about=” ht tp : //www. example . com/ bus ine s s /o r gan i s a t i on ”>

< r d f s : l a b e l>Organi sat ion</ r d f s : l a b e l><rdfs:comment>An organ i s a t i on /company</ rdfs:comment>

</ r d f s : C l a s s>

<rd f :P rope r ty rd f : about=” ht tp : //www. about . com/work”>< r d f s : l a b e l>work</ r d f s : l a b e l><rdfs:comment>The sub j e c t works somewhere</

rdfs:comment><r d f s : r a n g e r d f : r e s o u r c e=” ht tp : //www. example . com/

o rgan i s a t i on ”/><rd f s :domain r d f : r e s o u r c e=” ht tp : //www. example . com/

employee”/></ rd f :P rope r ty>

</rdf:RDF>

How RDF can be embedded into web data can be seen in Sec. A.2 (AppendixA).

OWL

The Web Ontology Language (OWL) is a knowledge representation languagedesigned by the W3C Web Ontology Working Group for use by applicationsthat need to process web content. It is a W3C recommendation since February2004 [11].

Page 12: Knowledge Representation Technologies in the Semantic Web

12

OWL was built on DAML+OIL, which is the abbreviation for Defence Ad-vanced Research Projects Agency (DARPA) Agent Markup Language + Ontol-ogy Inference Layer [12].

Contrary to XML, RDF and RDFS, OWL provides reasoning methods andadditional vocabulary: relations between classes, cardinality, equality, richer typ-ing of properties, characteristics of properties, and enumerated classes.

OWL Lite, OWL DL and OWL Full are the three levels of OWL.OWL Lite supports simple constraints and classification hierarchies which

are used to construct taxonomies and thesauri. The cardinality restriction islimited to 0 and 1.

OWL DL, where DL is the abbreviation for Description Logic, provides max-imum expressiveness while retaining computational completeness and decidabil-ity. Furthermore the cardinality between classes is not restricted to 0 and 1.

Unlike OWL DL, OWL Full does not guarantee computational completeness.Classes described in OWL Full can be treated simultaneously as instances ofcollections. It is implausible that a software will provide reasoning for everyOWL Full feature since complete OWL Full implementations do not currentlyexist. An advantage of OWL Full is that it is an extension to RDF while OWLLite and OWL DL are just an extension to restricted RDF. Hence, OWL Fullhas full RDF syntax support.

An OWL document can contain the following declarations:

– Header– Classes– Complex classes– Individuals– Properties, property characteristics and property restrictions– Ontology mapping

In the header span the ontology is defined, prior versions are mentioned,other OWL ontologies can be imported, a label can be assigned, comments canbe noted and other annotation properties can be predefined.

To arrange real world items we have to declare classes as in object orientedprogramming. It is also possible to build generalization/specialization relation-ships by defining subclasses with or without distinguished properties on thesuperclass.

Complex classes can be apportioned into set operators, enumerated classesand disjoint classes. OWL Lite does just allow intersectionOf as a set operatorand no enumerated or disjoint classes.

Individuals are the instances of classes.There exist two kinds of properties in OWL: Object Properties and Data-

type Properties. Object Properties define the aggregation between two classeswhereas Data-type Properties define the relation between a class and a literal.

Properties (exclusively Object Properties in OWL DL) can be characterizedby defining the type of the property as transitive, symmetric, functional, inverseand inverse functional. Table 1 shows the connection between these characteris-tics and the corresponding logical sentences.

Page 13: Knowledge Representation Technologies in the Semantic Web

13

Table 1: OWL property characteristics

Property characterisctic Logical sentence

TransitiveProperty P (x, y), P (y, z)⇒ P (x, z)SymmetricProperty P (x, y)⇔ P (y, x)FunctionalProperty P (x, y), P (x, z)⇒ y = zinverseOf ¬P (x, y) = P (y, x)InverseFunctionalProperty P (y, x), P (z, x)⇒ y = z

Property restrictions state about the amount of concerned properties andallow us to specify classes based on existing particular properties.

Defining equal classes and the equalities or differences between individuals iscalled ontology mapping. With property characteristics, restrictions and ontol-ogy mapping OWL provides rich semantics and the possibility to infer knowledgeand create a semi-complete entity world.

An OWL example code of an employee ontology can be seen in Listing 7.(Appendix A).

Read [13] for further information about the OWL Syntax.With XML/S, RDF/S and OWL we are able to build ontologies. In the next

section we will learn how to query the represented knowledge in order to scanthe data for terms of interest. Furthermore we will briefly have a look at ruleinterchanging methods.

4.3 Query Languages and Rules

Query LanguagesQuery Languages (QLs) are used to request data from data repositories.

In general, Semantic Web QLs can be divided into RDF-based QLs and OWLDL-based QLs.

RDF-based QLs proposals are more numerous as fish in the sea. The RDFData Query Language (RDQL) [14], the Second Generation RDF Query Lan-guage (SeRQL) [15] and SPARQL [16], which is a recursive acronym for SPARQLProtocol and RDF Query Language are a few examples.

Since SPARQL is a W3C recommendation since January 2008, we will brieflyget familiar with its concept. SPARQL has four query forms: SELECT, CONSTRUCT,ASK and DESCRIBE. The SELECT form is used to receive an object or literal ofan RDF triple. CONSTRUCT returns a template of a graph by substituting everyquery solution in the solution sequence and combining the triples into a singleRDF graph. ASK ascertains if a specific query pattern has a value and DESCRIBEreturns a description for a specific resource.

RDF-based data can be queried by defining specific constraints in the WHEREclause. Additional constraints can be achieved by adding a FILTER clause intothe WHERE clause for the use of regular expressions. The interested reader is askedto see the syntax specification [16].

Page 14: Knowledge Representation Technologies in the Semantic Web

14

Some of the OWL DL-based QLs are the Graphical query Language for OWLOntologies (GLOO) [17] and its extension the Visual Query Language for OWLOntologies (OntoVQL) [18], the Schema And Instance Query Language (SAIQL)[19], and SPARQL-DL [20], which is an extension to the RDF-based SPARQL.

SPARQL-DL is aligned to SPARQL and, therefore, provides interoperabilityof applications on the Semantic Web. It additionally provides query options forOWL DL property characteristics, restrictions and ontology mapping mecha-nisms.

Rules

Defining a set of rules is a natural form to encode knowledge. Common Logic(CL) [21], the Rule Markup Language (RuleML) [22] and the Rule InterchangeFormat (RIF) [23] are a few examples of efforts on building a high-level specifi-cation for exchanging rules in the Semantic Web.

The RIF W3C Working Group is currently producing W3C recommendationsfor rules interchange. It is developed to exchange rules between systems that useRDF/S, OWL Lite/OWL-DL or OWL Full. RIF is able to translate the rulesinto the internal rule-based language.

Logic unifies knowledge representation languages, Queries and Rules. Withthese technologies in use it becomes feasible to represent data, reason aboutthem and scan for data of interest. The vision of the Semantic Web will finallymaterialize.

5 Conclusion and outlook

In summary one can say that knowledge representation technologies are the corefield of the Semantic Web. Without having the possibility to store meta data ina structured way similar to our natural language it is not feasible to build on-tologies, which are important to correctly interpret the data in a communicationscene.

For ordinary websites this means that current keyword declarations will bereplaced by knowledge representing files. Therefore, search engines are able toobtain requested terms of specific contexts. The problem of ambiguity in wordsof natural languages is reduced.

Knowledge representing files can be accessed and scanned via query languageslike SPARQL which provides the presentation of requested terms of interest.

Furthermore, rules can be added to the data in order to infer further datarelations.

This will lead to a web for software agents, which are able to reason aboutthe data presented on websites and, thus, transaction processing will get morecomfortable due to inference automation.

The Semantic Web will be implemented and used by innovative organisationsthat attach importance to good knowledge management.

Page 15: Knowledge Representation Technologies in the Semantic Web

15

References

1. Das, S.: Role of semantic web in the changing context of digital environment.(2007)

2. Daconta, M., Obrst, L., Smith, K.: The Semantic Web. Wiley Publishing (2003)3. Thuraisingham, B.: Building Trustworthy Semantic Webs. Auerbach Publications

(2008)4. Berners-Lee, T.: The Semantic Web Layer Cake. http://www.w3.org/2007/03/

layerCake.svg(31.05.2008) (2007)5. ObjectManagementGroup: UML Resource Page. http://www.uml.

org/(02.06.2008) (2008)6. Russell, S., Norvig, P.: Artificial intelligence: a modern approach. (1995)7. Guarino, N., Giaretta, P.: Ontologies and knowledge bases: Towards a terminolog-

ical clarification. Towards Very Large Knowledge Bases (1995) 25–328. Manola, F., Miller, E.: Resource Description Framework (RDF):Concepts and

Abstract Syntax. http://www.w3.org/TR/2004/REC-rdf-concepts-20040210/

(22.05.2008) (2004)9. Beckett, D., McBride, B.: RDF/XML Syntax Specification (Revised). http://

www.w3.org/TR/rdf-syntax-grammar/ (24.05.2008) (2004)10. Manola, F., Miller, E.: RDF Primer. http://www.w3.org/TR/2004/

REC-rdf-primer-20040210/ (25.05.2008) (2004)11. McGuinness, D., Harmelen, F.: OWL Web Ontology Language Overview. http:

//www.w3.org/TR/owl-features/ (29.05.2008) (2004)12. Horrocks, I., Patel-Schneider, P., van Harmelen, F.: Reviewing the design of

DAML+ OIL: An ontology language for the semantic web. Proc. of the 18thNat. Conf. on Artificial Intelligence (AAAI 2002) (2002) 792–797

13. Smith, M., Welty, C., McGuinness, D.: OWL Web Ontology Language Guide.http://www.w3.org/TR/owl-guide/ (03.06.2008) (2004)

14. Seaborne, A.: RDQL - A Query Language for RDF. http://www.w3.org/

Submission/2004/SUBM-RDQL-20040109/ (06.06.2008) (2004)15. Broekstra, J., Kampman, A.: SeRQL: A Second Generation RDF Query Language.

SeRQL:ASecondGenerationRDFQueryLanguage (06.06.2008) (2003)16. Prudh́ommeaux, E., Seaborne, A.: SPARQL Query Language for RDF. http:

//www.w3.org/TR/rdf-sparql-query/ (05.06.2008) (2008)17. Fadhil, A., Haarslev, V.: GLOO: A Graphical Query Language for OWL Ontolo-

gies. OWL: Experience and Directions (2006) 215–26018. Fadhil, A., Haarslev, V.: OntoVQL: A Graphical Query Language for OWL On-

tologies. Proceedings of the 2007 International Workshop on Description Logics(DL-2007), Brixen-Bressanone, near Bozen-Bolzano, Italy, 810 June, 2007 (2007)267–274

19. Kubias, A., Schenk, S., Staab, S., Pan, J.: OWL SAIQLAn OWL DL QueryLanguage for Ontology Extraction. Proc. of OWLED 7 (2007)

20. Sirin, E., Parsia, B.: SPARQL-DL: SPARQL Query for OWL-DL. 3rd OWLExperiences and Directions Workshop (OWLED-2007) (2007)

21. Delugach, H.: Common Logic in Support of Metadata and Ontolo-gies. http://common-logic.org/docs/cl/Berlin OpenForum Delugach.pdf

(06.06.2008) (2005)22. Boley, H., Tabet, S., Wagner, G.: Design Rationale of RuleML: A Markup Language

for Semantic Web Rules. Proc. Semantic Web Working Symposium (SWWS01)(2001) 381–401

Page 16: Knowledge Representation Technologies in the Semantic Web

16

23. de Bruijn, J.: RIF RDF and OWL Compatibility. http://www.w3.org/2005/

rules/wiki/SWC (06.06.2008) (2008)

Page 17: Knowledge Representation Technologies in the Semantic Web

17

A Appendix

A.1 XML and XMLS

In Listing 3. we see an example of an XML code were two companies are describedby the elements name, type, url and location. To avoid the wrong interpretationof the words organisation, name and type we declare a namespace business witha URI and add the namespace to the equivocal elements. As we also want tovalidate our XML structure we have to link to the XMLS file xsd example.xsd.

The XMLS file, which acts as a template and a validation for the XMLdocument, can be seen in Listing 4.

Listing 3: XML example

<?xml version=” 1 .0 ” ?><xmlns :bus ine s s=” ht tp : //www. our−namespace−d e s c r i p t i o n s .

com/ bus ine s s ”><bu s i n e s s : l i s tO fO r g an i s a t i o n s xmlns :x s i=” ht tp : //www.w3 .

org /2001/XMLSchema−i n s t anc e ”xsi:noNamespaceSchemaLocation=”xsd example . xsd”><bu s i n e s s : o r g a n i s a t i o n>

<bus iness :name>Salzburg Research</ bus iness :name><bu s i n e s s : t yp e>co rpora t i on</ bu s i n e s s : t yp e><u r l>ht tp : //www. s a l zbu rg r e s e a r ch . at</ u r l>< l o c a t i o n>Salzburg , Austr ia</ l o c a t i o n>

</ bu s i n e s s : o r g a n i s a t i o n><bu s i n e s s : o r g a n i s a t i o n>

<bus iness :name>Sun microsystems</ bus iness :name><bu s i n e s s : t yp e>i n c o rpo ra t i on</ bu s i n e s s : t yp e><u r l>ht tp : //www. sun . com</ u r l>< l o c a t i o n l a t i t u d e=”37 2 1 ’N” long i tude=”121 5 8 ’W”>

Santa Clara , USA</ l o c a t i o n></ bu s i n e s s : o r g a n i s a t i o n>

</ bu s i n e s s : l i s tO fO r g an i s a t i o n s>

Listing 4: XMLS example

<?xml version=” 1 .0 ” ?><xs:schema xmlns :xs=” ht tp : //www.w3 . org /2001/XMLSchema”>

<xs : e l ement name=” l i s tO fOrgan i s a t i o n s ”><xs:complexType>

<xs : s equence><xs : e l ement r e f=” o rgan i s a t i on ” maxOccurs=”unbounded

”/></ xs : s equence>

</xs:complexType></ xs : e l ement>

Page 18: Knowledge Representation Technologies in the Semantic Web

18

<xs : e l ement name=” o rgan i s a t i on ”><xs:complexType>

<xs : s equence><xs : e l ement name=”name” type=” x s : s t r i n g ”/><xs : e l ement name=” type” type=” x s : s t r i n g ”/><xs : e l ement name=” ur l ” type=” x s : s t r i n g ”/><xs : e l ement name=” l o c a t i o n ” type=” x s : s t r i n g ”>

<x s : a t t r i b u t e name=” l a t i t u d e ” type=” x s : s t r i n g ”/>

<x s : a t t r i b u t e name=” long i tude ” type=” x s : s t r i n g ”/>

</xs:e lememt></ xs : s equence>

</xs:complexType></ xs : e l ement>

</xs:schema>

A.2 Embedding RDF

To offer software agents to obtain and reason about semantic web data we have toembed it into our web content. Listing 5 shows how an RDF/XML document canbe embedded into a HyperText Markup Language (HTML)/ eXtensible HTML(XHTML) web page. In Listing 6. we can see how RDF/XML is encoded into aScalar Vector Graphic (SVG) file.

Listing 5: RDF in HTML/XHTML

<head>< t i t l e>Henry ‘ s Website</ t i t l e><meta http−equiv=”Content−type ” content=’ t ext /html ;

cha r s e t=”utf−8” ’ />< l i n k r e l=” a l t e r n a t e ” type=” app l i c a t i on / rd f+xml” t i t l e=

”Henry ‘ s RDF data” h r e f=” rdf example . rd f ” /></head>

Listing 6: RDF in SVG

<svg><metadata><rdf:RDF xmlns : rd f=” ht tp : //www.w3 . org /1999/02/22− rdf−

syntax−ns#”xmlns:ab=” ht tp : //www. about . de/”xml:base=” ht tp : //www. henry . de/”>

<r d f :D e s c r i p t i o n rd f : ID=”Henry”ab:work=” ht tp : //www. job . org /”

Page 19: Knowledge Representation Technologies in the Semantic Web

19

ab:age=”23”><ab : f r i e nd rdf :nodeID=” s3 f o ” /></ r d f :D e s c r i p t i o n>

<r d f :D e s c r i p t i o n rdf :nodeID=” s3 f o ”ab:age=”23”>

</ r d f :D e s c r i p t i o n></rdf:RDF>

</metadata></ svg>

A.3 OWL example

The code example in Listing 7. demonstrates an employee ontology. An employeeis a subclass of a person with the minimum cardinality restriction of one on theproperty worksAt. The property worksAt forms a relation between an employeeand a company. A company has a name of the data type string and is locatedin a Region. locatedIn is a transitive property.

Listing 7: OWL example

<rdf:RDFxmlns : rd f=” ht tp : //www.w3 . org /1999/02/22− rdf−syntax−ns#”xmlns : rd f s=” ht tp : //www.w3 . org /2000/01/ rdf−schema#”xmlns:owl=” ht tp : //www.w3 . org /2002/07/ owl#”xmlns:xsd=” ht tp : //www.w3 . org /2001/XMLSchema#”xmlns:dc=” ht tp : //www.w3 . org /TR/2004/REC−owl−guide

−20040210/#DublinCore”>

< !−− Ontology header −−><owl:Ontology rd f : about=””>

<rdfs:comment>This onto logy d e s c r i b e s the world o f anemployee</ rdfs:comment>

<owl : import s r d f : r e s o u r c e=” ht tp : //www. example . org /per son onto logy ”/>

< r d f s : l a b e l>Employee Ontology</ r d f s : l a b e l><owl:AnnotaionProperty rd f : about=” d c : c r e a t o r ”/>

</ owl:Ontology>

< !−− Cla s s : Employee −−>< !−− A sub c l a s s o f Person t ha t works at l e a s t a t one

company −−><ow l :C la s s rd f : ID=”Employee”>

<rd f s : subC la s sO f r d f : r e s o u r c e=”#Person”><ow l :R e s t r i c t i o n>

<owl :onProperty r d f : r e s o u r c e=”#worksAt”/>

Page 20: Knowledge Representation Technologies in the Semantic Web

20

<owl :minCard ina l i ty rd f : da ta type=”&xsd ;nonNegat iveInteger ”>1</ owl :minCard ina l i ty>

</ ow l :R e s t r i c t i o n></ rd f s : subC la s sO f><d c : c r e a t o r>Stephanie Stroka</ d c : c r e a t o r>

</ owl :C la s s>

< !−− Cla s s : Company −−><ow l :C la s s rd f : ID=”Company”>

<d c : c r e a t o r>Stephanie Stroka</ d c : c r e a t o r></ owl :C la s s>

< !−− Cla s s : Region −−><ow l :C la s s rd f : ID=”Region”>

<d c : c r e a t o r>Stephanie Stroka</ d c : c r e a t o r></ owl :C la s s>

< !−− Employee works at Company −−><owl :ObjectProperty rd f : ID=”worksAt”>

<rd f s :domain r d f : r e s o u r c e=”#Employee”/><r d f s : r a n g e r d f : r e s o u r c e=”#Company”/><d c : c r e a t o r>Stephanie Stroka</ d c : c r e a t o r>

</ owl :ObjectProperty>

< !−− Company has company name o f da t a t y p e : s t r i n g −−><owl:DatatypeProperty rd f : ID=”companyName”>

<rd f s :domain r d f : r e s o u r c e=”#Company”/><r d f s : r a n g e r d f : r e s o u r c e=”&xsd ; s t r i n g ”/><d c : c r e a t o r>Stephanie Stroka</ d c : c r e a t o r>

</ owl:DatatypeProperty>

< !−− Everyth ing i s l o c a t e d in a Region −−>< !−− Trans i t i v e Proper ty : P( x , y ) , P(y , z ) => P(x , z ) −−><owl :ObjectProperty rd f : ID=” lo ca t ed In ”>

<r d f : t yp e r d f : r e s o u r c e=”&owl ; Trans i t i veProper ty ”/><rd f s :domain r d f : r e s o u r c e=”&owl ; Thing”/><r d f s : r a n g e r d f : r e s o u r c e=”#Region”/><d c : c r e a t o r>Stephanie Stroka</ d c : c r e a t o r>

</ owl :ObjectProperty>

< !−− I n d i v i d u a l : Region=Austr ia −−><Region rd f : ID=” Austr ia ”/>

< !−− I n d i v i d u a l : Region=Sa l z burg −−><Region rd f : ID=”Salzburg ”>

Page 21: Knowledge Representation Technologies in the Semantic Web

21

<l o c a t ed In r d f : r e s o u r c e=”#Austr ia ”/></Region>

< !−− I n d i v i d u a l : Region=Puch −−><Region rd f : ID=”Puch”>

<l o c a t ed In r d f : r e s o u r c e=”#Austr ia ”/></Region>

< !−− I n d i v i d u a l : Company=Sa l z burg Research −−>< !−− l o c a t e d in Sa lzburg , t h e r e f o r e a l s o l o c a t e d in

Austr ia ( Trans i t i v e Property ) −−><Company rd f : about=” ht tp : //www. s a l zbu rg r e s e a r ch . at ”>

<companyName rd f : da ta type=”&xsd ; s t r i n g ”>SalzburgResearch</companyName>

<l o c a t ed In r d f : r e s o u r c e=”#Salzburg ”/></Company>

< !−− I n d i v i d u a l : Company=FH Sa l z burg −−>< !−− l o c a t e d in Puch , t h e r e f o r e a l s o l o c a t e d in Austr ia (

Trans i t i v e Property ) −−><Company rd f : about=” ht tp : //www. fh−sa l zburg . ac . at ”>

<companyName rd f : da ta type=”&xsd ; s t r i n g ”>FachhochschuleSalzburg</companyName>

<l o c a t ed In r d f : r e s o u r c e=”#Puch”/></Company>

< !−− I n d i v i d u a l : Company=Employee −−><Employee rd f : ID=”Dietmar Glachs”>

<worksAt r d f : r e s o u r c e=” ht tp : //www. s a l zbu rg r e s e a r ch . at ”/>

<worksAt r d f : r e s o u r c e=” ht tp : //www. fh−sa l zburg . ac . at ”/></Employee>

</rdf:RDF>