Game Development Reference
In-Depth Information
their use. In our work, we recognize two major types: the core semantics (domain
models) and resource descriptions (also referred as annotations or resourcemetadata).
The core of the Semantic Web are the ontologies, which model the domain knowl-
edge : the common facts of the world (e.g., “carnivorous animals eat animals”),
abstract or concrete concept definitions (e.g., “mammal is an animal”) or constraints
and rules of the world (e.g., “mammals have between 2 or 4 legs”). Ontologies are
the world abstraction and represent an worldwide agreement on domain's general
rules.
Ontologies that are highly structured and elaborated are often called the “heavy-
weight” semantics. Though they provide advanced capabilities, they are also hard
to obtain and are almost exclusively built by (costly) human experts. The cheaper
and more easy-to-create are the “lightweight” semantics with “lightweight” rep-
resentations. These include taxonomies (hierarchical organizations of entities) and
free association networks (which express general relatedness of entities) sometimes
referred to as folksonomies (in case they were created using a crowdsourcing tech-
nique). The semantic structures sometimes use simple terms instead of concepts (also
due to ease of creation), which of course, brings drawbacks like semantic ambiguity.
The prevailing standards for ontology representation are RDF (resource descrip-
tion framework) and OWL (web ontology language). The basic elements of ontolo-
gies are atomic facts expressed in the form of triplets consisting of the source and
target entity ( subject and object ) connected with a predicate (e.g., “dog” (subject)—
“is a” (predicate)—“mammal” (object)). Triplets represent various types of relation-
ships among entities (e.g., hierarchy, composition, usage). Each entity or predicate
has a textual description, but also URI that identifies it globally. The ontology entities
represent concepts (classes) or their instances. Each entity can also be decorated with
literal properties. Using these basic elements, more complex facts are composed.
As second part of the Semantic Web, resource descriptions are the connection
between the domain models and web resources. For example, they may denote to
which particular concepts the resources are related (e.g., “this article is about bears”)
or they provide additional structural information about particular document (e.g.,
document outline based on semantic properties of individual paragraphs).
Resource annotations are represented either in the knowledge bases (e.g., as RDF
triplets) or as a direct part of the web resources itself (e.g., meta-tags of HTML,
RDFa). They also vary in terms of what kind of resource (or its part) they annotate.
Different kinds of annotations are found in case of texts (the annotation can be related
to whole text, paragraph, sentence or even single word), images (spatial information),
audio tracks or videos (temporal information).
2.2 Semantics in Use
The web semantics is utilized in various applications. Typically, they are connected
to the information retrieval, information space organization, navigation or recom-
mender systems.
 
Search WWH ::




Custom Search