RAZUMEVANJE IN REPREZENTACIJA ZNANJA

Abstract

Long-term memory is the depository for our relatively stable knowledge and skills, including knowledge of language, spatial models of the world, knowledge of the properties of objects and people, events in our own lives, sensorimotor skills, etc. There exist numerous formalisations for representing knowledge, but associative (or semantic) networks and production rule systems are the most commonly used. In associative (or semantic) networks, the basic elements of memory are concepts (knots, symbols), and relations between concepts (propositions, symbolic structure). A concept may be perception-based, action-based, a basic logical or semantic relation or a higher-level concept built on relations between such elements. Concepts may be general terms or individual constants. The meaning of a concept is determined in part by the configuration of its relations to other concepts, and in part the referential conditions that are essential for the correct use and application of a term. Production is a rule in the form «if-then», which states what kind of operation must be executed when certain conditions apply. The conditions may be any pattern of activity represented in the long-term memory as a complex proposition; actions may be external or internal moves such as searching for a memory location, seeking a referent in the memory, or activating other long-term memory concepts.