Deep Lexical Semantics: The Ontological Ascent

Deep Lexical Semantics: The Ontological Ascent∗ Jerry R. Hobbs Information Sciences Institute, University of Southern California Marina del Rey, California, USA, hobbs@isi.edu Abstract. Concepts of greater and greater complexity can be constructed by building systems of entities, by relating other entities to that system with a figure-ground relation, by embedding concepts of figure-ground in the concept of change, by embedding that in causality, and by coarsening the granularity and beginning the process over again. This process can be called the Ontological Ascent. It pervades natural language discourse, and suggests that to do lexical semantics properly, we must carefully axiomatize abstract theories of systems of entities, the figure-ground relation, change, causality, and granularity. In this paper, I outline what these theories should look like. 1. Introduction If we are going to have programs that understand language, we will have to encode what words mean. Since words refer to the world, their definitions will have to be in terms of some underlying theory of the world. We will therefore have to construct that theory, and do so in a way that reflects the ontology that is implicit in natural language. There are wrong ways to go about this enterprise. For example, we could take our underlying theory to be quantum mechanics and attempt to define, say, verbs of motion in terms of the primi- tives provided by that theory. A less obviously wrong approach, and one that has sometimes been tried, is to adopt Euclidean 3-space as the underlying model of space and attempt to define, say, spatial prepostions in terms of that. In this paper, I propose a general structure for a different underlying conceptualization of the world—one that should be particularly well suited to language. It consists of a set of core theories of a very abstract character. These theories are too abstract to impose many constraints on the entities and situations they are applied to. In fact, the reader may complain that they apply to anything. But the main purpose of the core theories is to provide the basis for a rich vocabulary for talking about entities and situations. The fact that the core theories apply so widely means that they provide a great many domains of discourse with a rich vocabulary. These will not be just any core theories, but theories that find their place in a schema I call the “Ontological Ascent”. They include a theory of systems or composite entities and the figure- ground relation, which subsumes a theory of scales; a theory of change of state; a theory of causality, which provides support for a theory of goal-directed behavior; and very importantly a theory of shifts in granularity. The Ontological Ascent is described in Section 2. Then in the subsequent sections I sketch the outlines of each of the above theories. The enterprise is therefore to axiomatize these core theories in as clean a fashion as possible, and then to define, or at least characterize, various words in terms of predicates supplied by these core …

Content truncated to 500 words