The Theoretical Status of Ontologies in Natural Language Processing

TL;DR

This paper reviews the role of ontologies in NLP, classifies different types, discusses benefits and challenges, and shares insights from using the Upper Model, proposing future design criteria.

Contribution

It provides a comprehensive classification of NLP ontologies, analyzes the Upper Model's application, and suggests new criteria for NLP ontology design.

Findings

The Upper Model has been effectively used in text generation.

Different types of ontologies offer various benefits and face specific challenges.

Proposed criteria aim to improve future NLP ontology development.

Abstract

This paper discusses the use of `ontologies' in Natural Language Processing. It classifies various kinds of ontologies that have been employed in NLP and discusses various benefits and problems with those designs. Particular focus is then placed on experiences gained in the use of the Upper Model, a linguistically-motivated `ontology' originally designed for use with the Penman text generation system. Some proposals for further NLP ontology design criteria are then made.