Description Logics

Abstract

Description logics¹ are languages tailored for expressing knowledge about concepts and concept hierarchies. The concept hierarchies represent the is-a relation between concepts. Description logics are usually given a Tarski-style declarative semantics, which allows them to be seen as sub-languages of predicate logic.² The main entities in description logics are

• concepts, which can be considered as unary predicates which are interpreted as sets of objects over a domain. Examples of concepts are soccer-team and ida-employee.³

• roles which can be considered as binary predicates which are interpreted as binary relations between the objects in the domain. An example of a role is member which may represent a relation between a team and the persons belonging to the team.

• individuals which are interpreted as objects in the domain. For example, a particular member of a team would be represented by an individual.

To build a knowledge base one starts with primitive concepts and roles (see section 2.1.1 for the definition), and can use the language constructs (such as intersection, union or role quantification) to define new concepts and roles. Also information about individuals can be told to the knowledge base.

Description logics have also been referred to as terminological logics, concept languages and KL-ONE-like languages. They have their origin in semantic networks and frame-based systems. The web page of the description logic community is found at address http://dl.kr.org/dl.

See, for instance, [Bor94] for a formal discussion.

IDA is the Swedish name for the department of computer and information science.