Building Large Knowledge-Based Systems; Representation and Inference in the Cyc Project
Building Large Knowledge-Based Systems; Representation and Inference in the Cyc Project
| Hi-index | 0.00 |
To achieve a human level of understanding of natural language, cognitive systems need a representation of knowledge that can distinguish nuances of meaning as well as preserve intended ambiguity. Among the most important of the basic concepts in such a knowledge base are the representations for events and processes, and also for objects and substances. Although each concept in these related pairs of concepts appears distinct intuitively and linguistically, the distinction is sometimes lost when the concepts are formalized. In different ontologies one finds a variety of different representations for these four fundamental concepts. This paper presents a new representation which uses an extension of dimensional analysis to qualitative dimensions to provide distinct and logically well-defined yet intuitively acceptable definitions for these four fundamental concepts, and specifies how they relate to other fundamental concepts in an upper ontology. In this analysis, processes extended through time result in events, and substances extending through space constitute objects. These representations are expected to be better suited than alternatives for representing the nuances of meaning in linguistic utterances.