15 - The Omega ontology

Summary

Introduction

We present the Omega ontology, a 120,000-node terminological ontology constructed at USC ISI as the synthesis of WordNet 2.0 (Miller, 1990; Fellbaum, 1998), a lexically oriented network constructed on general cognitive principles, and Mikrokosmos (Mahesh, 1996; O'Hara et al., 1998), a conceptual resource originally conceived to support translation, whose result is subordinated under a new feature-oriented upper model, created expressly in order to facilitate the merging of lower models into a functional whole.

Omega, like its close predecessor SENSUS (Knight and Luk, 1994b), can be characterized as a shallow, lexically oriented, term taxonomy – by far the majority of its concepts can be stated in English using a single word. At present, Omega contains no formal concept definitions and only relatively few interconnections (semantic relations) between concepts, in contrast to work such as DOLCE (Chapter 3) and SUMO (Chapter 2). Besides the core concept base, Omega has been extended to connect with a range of auxiliary knowledge sources (including instances, verb-frame annotations, and domain-specific sub-ontologies) incorporated into the basic conceptual structure and representation. By making few commitments to any specific theories of semantics or particular representations, Omega enjoys a malleability that has allowed it to be used in a variety of applications, including question answering and information integration.

We describe ongoing work on the manual reorganization of Omega terms and their connection to a large corpus of manually disambiguated word senses in three languages.