The Semantic Web - Enabling Semantic Web Services: The Web Service Modeling Ontology - page 28

• Informal taxonomy. There is an explicit hierarchy (generalization and spe-

cialization are supported), but there is no strict inheritance; an instance of
a subclass is not necessarily also an instance of the superclass. An example
is the Yahoo! Dictionary. 4

• Formal taxonomy. There is strict inheritance; each instance of a subclass

is also an instance of a superclass. An example is UNSPSC.

• Frames. Frame (or class) contains a number of properties and these prop-

erties are inherited by subclasses and instances. Ontologies expressed in
RDFS [20], described below, fall into this category.

• Value restrictions. The values of properties are restricted. Ontologies ex-

pressed in OWL Lite (see Section 3.3) fall in this category.

• General logic constraints. Values may be constrained by logical or mathe-

matical formulas using values from other properties. Ontologies expressed
in OWL DL (see Section 3.3) fall into this category.

• Expressive logic constraints. Very expressive ontology languages such as

those seen in Ontolingua [36] and CycL [82] allow first-order logic con-
straints between terms and more detailed relationships such as disjoint
classes, disjoint coverings, inverse relationships, and part-whole relation-
ships. Note that some of these detailed relationships, such as disjointness
of classes, are also supported by OWL DL (and even OWL Lite), which in-
dicates that the borders between the levels of expressiveness remain fuzzy.

3.1.4 History of Ontology Languages

In the areas of knowledge engineering and knowledge representation, interest
in ontologies really started taking off in the 1980s with knowledge represen-
tation systems such as KL-ONE [19] and CLASSIC [18].
An important system for the development, management, and exchange of
ontologies in the beginning of the 1990s was Ontolingua [36], which uses an
internal KIF (knowledge interchange format) 5 representation, but is able to
interoperate with many other knowledge representation (ontology) languages,
such as KL-ONE, LOOM, and CLASSIC.
The languages used for ontologies were determined by the tool used to cre-
ate the ontologies. Systems such as KL-ONE, CLASSIC and LOOM each used
their own ontology language, although the Ontolingua system was capable of
translating ontologies between different languages, using the KIF language as
an interchange language. We can see the languages and tools as being interde-
pendent, but also as being somewhat orthogonal, where we have the language
on one axis and the tool on the other. For example, KL-ONE, CLASSIC, and
LOOM all have their basis in description logics [5], while KIF has its basis in
first-order logic.

4 http://www.yahoo.com .
5 http://logic.stanford.edu/kif/kif.html .