| Semantic Web is the trend for the next stage of the Web, while ontology is the"core level" which describes semantic relations between concepts in the semanticWeb system. At present there are a lot of studies on ontology, and experts fromvarious fields are constructing domain ontologies. But some problems still exist inthe process of ontology construction: the first is the quality problem. Constructionof ontology costs a great many labor and resources to collect the relations betweenvarious concepts in certain domain, therefore, some cases are very possible tohappen due to man-made mistakes such as inadequate concepts in the domain andinexact conceptual relation definition affecting the quality of ontology. The secondis the inconsistency in the concept relation description. Now there is no unifiedknowledge organization system for the construction of domain ontologies, anddifferent labels are being used to describe conceptual relations in variousontologies, leading to the condition that interoperation can not be conducted amongontologies and reducing the reusability of ontologies.Knowledge Organization Systems (KOS) is the terminology and sign systemthat we use to define and organize and represent real-world objects, generallyrefereed to be semantic tools in concrete application, including the thesaurus inlarge-scale database, classification schemes and automatic retrieval expansionvocabularies inside the search engine, grading systems for Website navigation andknowledge ontology of semantic Web. Knowledge Organization Systems exertimportant effect in many digital libraries and digital classification systems.Traditional knowledge organization tools thesaurus and ontology are used todescribe conceptual relations similarly: they both choose concepts and terms fromtheir own domains, with classification schedules both being grading structurecontaining annotation information;and they both choose some identifiers torepresent conceptual relations. These similarities are the key issues in domainontology construction. Because of the natural relation between thesaurus andontology in representing knowledge structure, many academic groups overseas anddomestically attempted to construct ontology with existing thesauri in successionafter the semantic Web was proposed. Up to now there has been 10-odd thesauribeing transformed to ontologies with various methods.There is no unified knowledge organization system standard for thetransformation from thesaurus to ontology. Oversea studies focus on therepresentation of thesaurus with Resource Description Framework SchemaLanguage (RDF Schema). RDF is characterized by flexibility and expansion, easyto combine with other resource organization systems. In this regard, it is suitablefor the transformation from thesaurus to ontology. However, RDF Schema onlyprovides primary semantic relation representation, without unified labels to supportsemantic relation description which is more concrete. Therefore, it is necessary toestablish the knowledge organization system which is flexible and expandable tosupport more concrete semantic relations on the basis of RDF.In 2005, W3C recommended Simple Knowledge Organisation System(SKOS), which is based on RDF Schema and expands it, providing special labels todescribe and share controlled thesaurus. The objective of SKOS is to provide apowerful framework of knowledge representation in a machine-understandable wayfor knowledge organization of semantic Web. There has been some precisevocabulary forms expressed with XML and RDF heretofore. RDF vocabulary cameinto being in the DESIRE project, and was developed further in the Limber project.All the work laid a foundation for SKOS Core. The earliest application of SKOS inthe semantic Web is in a sub-project "Catalogue and Thesaurus management" ofSWAD-E. Now, the RDF vocabulary schema draft compatible with ISO relatedstandards further develops map and navigation tools among vocabularies.SKOS Core provides a framework by connecting the concept with thevocabularies and phrases commonly used by people. SKOS Core is a model forexpressing the basic structure and content of knowledge organization system,providing a series of RDF properties and RDF(S) Labels describing the basicstructure and content of concept schemes (thesauri, classification schemes, subjectheading lists, taxonomies, 'folksonomies' and other types of controlled vocabulary),including Concept,prefLabel,altLabel,broader,narrower,scopeNote and so on.SKOS provides a standard framework of knowledge organization for thetransformation from thesaurus to ontology, under the SKOS standard, unified labelsand descriptions are used in the transformation, creating conditions for the reuseand interoperation of ontology after transformation.The author chose one paragraph of UK Archive Thesaurus (UKAT), andconstructed an ontology based on the SKOS standard with the ontology edition toolProtégé. This is also the emphasis of the paper.Construction of ontology may be divided into the following steps:Step 1: To introduce Namespace. Namespace is the precondition to use aseries of items in ontology, and we must correctly explain the vocabulary that isbeing used. The standard initial module of ontology is a statement of a series ofNamespaces that is contained in rdf:RDF labels. These labels are used to explainthe identifiers in files correctly, so as to make other parts of ontology havereadability.Step 2: to construct class. Class is also called concepts, expressing theaggregation of objects. It is the most basic and important modeling primitive, is thefoundation to construct ontology. In SKOS-based ontology construction, twoclasses expressing scheme and concept in SKOS are used in the construction ofclass: skos:ConceptScheme and skos:Concept. Of which, skos:ConceptScheme isthe chosen scheme class, provides corresponding descriptive metadata informationfor skos:ConceptScheme, explaining name of chosen scheme (dc:title) andchosen paragraph (dc:description), and so on. Skos:Concept is the concept class ofscheme. All concepts of chosen paragraphs in the vocabulary are the narrowerclasses of skos:Concept. Then, corresponding class system is built according to therelations of grading trees of concepts in chosen paragraphs of the vocabulary.Step 3: to construct property. Property can be used to explain the commoncharacteristics of classes and special characteristics of some objects. A property is abinary relation, and it can be restricted by appointing the domain and range ofproperty or defining sub-property to express relations of objects in a range morecorrectly. Property also has other restrictions, such as TransitiveProperty ,SymmetricProperty,hasValue,Cardinality. Majority of the labels in SKOS Coreappear in the form of property, and the main properties involved in this case are asfollows: skos:prefLabel , skos:scopeNote , skos:inScheme , skos:broader ,skos:narrower , skos:related , skos:altLabel. They respectively correspond topreferred identifier, range annotation, affiliated vocabulary, wider or narrowerconcepts, related items, various grades like replacement items, correlativity.Step 4: to add property for concepts. Every concept corresponds to someproperties, which should be added and given corresponding values according to therelations in vocabularies.We can see from the process of ontology construction that, SKOS expressesthe structure and content of vocabulary on the basis of RDF, it is simpler in logicthan other ontology description languages for precise reasoning, and needs no morerestrictions. Emphases of vocabulary ontology construction lay on differentiatingclass, property, relations and grasping crossing relation. The ontology author builtis just a primary attempt, and more work still needs to be done in order to realizecomplete ontology construction with vocabulary expressed by SKOS. |
PDF Full Text Request