Catalogue of Artificial Intelligence Techniques

   

Jump to: Top | Entry | References | Comments

View Maths as: Images | MathML

Conceptual Graphs

Aliases: CGs

Keywords: Existential Graphs, entity-relationship, type hierarchy

Categories: Knowledge Representation , Natural Language


Author(s): Nicolas Nicolov

Conceptual Graphs (CGs) are a knowledge representation formalism, a variant of Semantic Networks. CGs were developed by John Sowa and stem from the Existential Graphs of Charles Sanders Peirce which are a graphic notation for Classical Logic. A conceptual graph is a structure of concepts and conceptual relations linked with arcs where every arc connects a conceptual relation to a concept. The types of the concepts and the relations form concept- and relation-hierarchies (lattices). A graph can be embedded in another graph by means of the context mechanism. The default existential quantifier and the possibility to negate contexts give CGs an equivalent power to classical predicate calculus. CGs have definitional mechanisms that can support extensions to the core in a controlled, systematic way. Because CGs can be mapped to classical predicate calculus (or order-sorted logic), they are thus seen as a (graphical) notation for logic. However, it is the topological nature of formulas which CGs make clear, and which can be exploited in reasoning and processing. CGs are intuitive because they allow humans to exploit their powerful pattern matching abilities to a larger extent than does the classical notation--thus, CGs are particularly useful for the interchange of knowledge between humans and computers. CGs can be viewed as an attempt to build a unified modelling language and reasoning tool. They can model data, functional and dynamic aspects of systems. They form a unified diagrammatic tool which can integrate entity-relationship diagrams, finite-state machines, Petri nets, and dataflow diagrams. CGs have found wide application in systems for information retrieval, database design, expert systems, conceptual modelling and natural language processing. See Graph Matching.


References:


Comments:

Add Comment

No comments.