Issues of Semantics in Semantic-Network Representation of Belief
Robin K. Hill · 1994
Knowledge representation and reasoning systems that are used for cognitive modeling must capture mental intensions, i.e., senses of linguistic or other semantic constructions, as opposed to (or, rather, in addition to) references, such as physical objects, of those constructions. A semanticnetwork system used for this purpose needs a semantics in which the nodes of a network are terms in the language of thought of the modeled cognitive agent and represent entities in the agent's mental universe. SNePS ("Semantic Network Processing System") is such a system. With SNePS as the framework, specific issues studied include how different types of nodes (concepts) are to be interpreted and used by the cognitive agent and the different types of computations required to support these uses; how nodes come to be distinct or similar, from both inter-agent and intra-agent perspectives; how (and why) circularity of meaning may be accommodated in a semantic network; and to what extent the coherence of...