A cognitive perspective on knowledge representation

Abstract

Knowledge representation is at the heart of every artificial intelligence system. The issues involved in the design of a representation are not restricted to technical aspects of the representation formalism (storage, access, inference, assimilation, consistency), but include the modeling process (relevance and granularity issues). We propose an extended knowledge representation model, characterized by making the role of the observer explicit. We use the representation model to look into the various modalities of representation such as declarative, procedural, prepositional, analogical, etc. Generally only some aspects of a representation correspond to a particular modality, depending on the level of abstraction considered. The concept of qualitativeness is found to be orthogonal to the modalities discussed.