Ontology and the Lexicon

Introduction

Combining large lexical resources with world knowledge, via ontologies, is a crucial step for reasoning over natural language, particularly for the Semantic Web. Concrete NLP applications include semantic parsing, text summarization, translation, and question answering. For example, questions like ‘Could Y have murdered X?’ require several inference steps based on semantic facts that simple lexicons do not include. An essential ingredient is so-called openworld semantics offered by state-of-the-art Description Logic (DL) reasoners, e.g., FaCT (Horrocks, 1998) or Racer (Wessel and Möller, 2005).

The FrameNet lexicon (Ruppenhofer et al., 2006a) has a uniquely rich level of semantic detail, which is based on frame semantics (and not on open-world semantics). Therefore, we are building bindings from FrameNet to multiple ontologies. For a specific application, we choose an appropriate ontology and build the binding for this application. Thus, we enable reasoners to make inferences over natural language text.

In this chapter, we report on the first steps towards this goal: we have automatically translated a crucial portion of FrameNet to OWL DL, and we show how state-of-the-art DL reasoners can make inferences over FrameNet-annotated sentences. Thus, annotated text becomes available to the Semantic Web, and FrameNet itself can be linked to other ontologies. While our OWL translation is limited to facts included in FrameNet, links to ontologies make world knowledge available to reasoning over natural language text. Therefore, we link FrameNet to the Suggested Upper Merged Ontology (SUMO).