Searching Semantic Resources for Complex Selectional Restictions to Support Lexical Acquisition

Abstract

Natural language processing systems are increasingly using ontologies and other large-scale semantic resources to support Verb Sense Disambiguation (VSD) and other applications. One of the ways in which these resources can be used is to identify the selectional restrictions on verb arguments needed for sense distinction. However, manually navigating such resources can be difficult and inefficient due to their size and complexity. In this paper, we present a process for automatically searching through an ontology to determine appropriate concepts for expressing selectional restrictions on verb sense. The goal of this research is to semi-automate the development of a semantically rich lexicon to support high-precision information extraction.