Key concepts in the semantic web
| Key concept | Description |
|---|---|
| Resource | Anything published under the Semantic Web vision, that is, with links to other resources. For example, a dataset on influenza or the concept of Europe |
| Concept | A kind of resource that represents a single idea. For example, the concept of influenza |
| Identifier | A unique string of characters that refers to exactly one concept. For instance, ‘http://purl.obolibrary.org/obo/DOID_8469’, which identifies the concept of influenza |
| Ontology | A collection of concepts and the relationships between them. These relationships provide concepts with a machine-readable meaning that can be explored for pattern recognition, knowledge discovery or any other computational analysis. Ontologies are the main source of concepts used in the metadata of Semantic Web resources. |
| Metadata | A machine-readable description of the contents of a resource made through linking the resource to the concepts that describe it. For instance, a dataset links to the concept of ‘influenza’ because it contains data concerning that disease. Figure 1 illustrates the metadata of a resource |
| Annotation | The process of enriching a resource with information about itself (metadata) by means of semantically defined properties pointing to other resources, especially properties pointing to concepts from ontologies |
| Property | A relationship between two resources, such as the ones that are used to annotate a resource with its metadata. Properties are also often called relations. |
| Semantic search engine | A software that searches for annotated resources in the Semantic Web based on a query. Inference can be used to find the appropriate results; additionally, results can be ranked according to similarity and relevance to the query. See the text for an example |
| Pattern recognition | The application of ontology-stored knowledge and inference mechanisms in the pursuit of unapparent correlations found in data. By using the semantics of the resources, information that is not explicit in the data but can be derived from ontologies can lead to more comprehensive results (see the text for an example). |