370
M. Weber and E. A. Lee
1.3 Semantic Technologies
The term “semantic technology” describes a collection of popular standards and
technologies for representing and working with ontologies, be they spatial or
otherwise.
The Semantic Web was proposed by Tim Berners-Lee in 2001 as an extension of
the World Wide Web that would allow ordinary HTML web pages to be enhanced
with special markup to label their semantic content. The hope is that when markup
is combined with a collection of ontologies for web content and real-world objects,
algorithms will be able to apply ontological reasoning to web elements and data. For
example, an image of a bridge embedded in a web site could be labeled as such and
found through a general search for “landmarks” by using the ontological information
that a bridge is a landmark. According to the wikipedia article on the semantic web, by
2013 some 4 million web pages had been augmented with semantic web information.
But this is done primarily through human intervention, which could account for the
relatively modest penetration compared to the total number of web pages.
Semantic Web ontologies are expressed in Resource Description Framework
(RDF), an abstract model for semantic data as sentence-like statements about the
world in triples of subject, predicate, object. For example: the sentence “A cow”
(subject) “isa” (predicate) “farm animal” (object), or “The mall parking lot” (subject)
“has the number of free spaces” (predicate) “45” (object). As hinted at by these
examples, triples can express both abstract information about classes (cows and
farm animals) as well as facts about specific instances (the mall parking lot) and raw
data values (45). A database designed and optimized for RDF data is known as a
semantic repository or alternatively a triple store. The W3C SPARQL Protocol and
RDF Query Language (SPARQL) recommendation [12] defines both a protocol and
a query language for performing SQL-like operations on a semantic repository such
as queries, inserts, updates, and deletes.
RDF is a natural way to express relational ontologies, as discussed in Sect. 2.3,
for semantic localization. Additionally, some semantic repositories, like GraphDB,
support geospatial plugins for efficient queries over geocoordinates (i.e. latitude and
longitude pairs). If compatible with the GeoSPARQL standard [13], the semantic
repository may also be able to automatically derive RCC (Region Connection
Calculus) relationships, such as containment of one geospatial object within another,
directly from the definitions of the objects themselves.
1.4 Standards for Spatial Representation
Many standards for spatial representation have been proposed in different domains,
a sample of which is presented here.
According to Lieberman et al., as of 2007 the semantic web maintained at least
seven varieties of spatial ontologies [14]. These include Geospatial Features, Feature
M. Weber and E. A. Lee
1.3 Semantic Technologies
The term “semantic technology” describes a collection of popular standards and
technologies for representing and working with ontologies, be they spatial or
otherwise.
The Semantic Web was proposed by Tim Berners-Lee in 2001 as an extension of
the World Wide Web that would allow ordinary HTML web pages to be enhanced
with special markup to label their semantic content. The hope is that when markup
is combined with a collection of ontologies for web content and real-world objects,
algorithms will be able to apply ontological reasoning to web elements and data. For
example, an image of a bridge embedded in a web site could be labeled as such and
found through a general search for “landmarks” by using the ontological information
that a bridge is a landmark. According to the wikipedia article on the semantic web, by
2013 some 4 million web pages had been augmented with semantic web information.
But this is done primarily through human intervention, which could account for the
relatively modest penetration compared to the total number of web pages.
Semantic Web ontologies are expressed in Resource Description Framework
(RDF), an abstract model for semantic data as sentence-like statements about the
world in triples of subject, predicate, object. For example: the sentence “A cow”
(subject) “isa” (predicate) “farm animal” (object), or “The mall parking lot” (subject)
“has the number of free spaces” (predicate) “45” (object). As hinted at by these
examples, triples can express both abstract information about classes (cows and
farm animals) as well as facts about specific instances (the mall parking lot) and raw
data values (45). A database designed and optimized for RDF data is known as a
semantic repository or alternatively a triple store. The W3C SPARQL Protocol and
RDF Query Language (SPARQL) recommendation [12] defines both a protocol and
a query language for performing SQL-like operations on a semantic repository such
as queries, inserts, updates, and deletes.
RDF is a natural way to express relational ontologies, as discussed in Sect. 2.3,
for semantic localization. Additionally, some semantic repositories, like GraphDB,
support geospatial plugins for efficient queries over geocoordinates (i.e. latitude and
longitude pairs). If compatible with the GeoSPARQL standard [13], the semantic
repository may also be able to automatically derive RCC (Region Connection
Calculus) relationships, such as containment of one geospatial object within another,
directly from the definitions of the objects themselves.
1.4 Standards for Spatial Representation
Many standards for spatial representation have been proposed in different domains,
a sample of which is presented here.
According to Lieberman et al., as of 2007 the semantic web maintained at least
seven varieties of spatial ontologies [14]. These include Geospatial Features, Feature