ONTOGRABBING: Extracting Information from Texts Using Generative Ontologies

  • Authors:
  • Jørgen Fischer Nilsson;Bartłomiej Antoni Szymczak;Per Anker Jensen

  • Affiliations:
  • DTU Informatics, Technical University of Denmark,;DTU Informatics, Technical University of Denmark, and Copenhagen Business School, International Language Studies and Computational Linguistics,;Copenhagen Business School, International Language Studies and Computational Linguistics,

  • Venue:
  • FQAS '09 Proceedings of the 8th International Conference on Flexible Query Answering Systems
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

We describe principles for extracting information from texts using a so-called generative ontology in combination with syntactic analysis. Generative ontologies are introduced as semantic domains for natural language phrases. Generative ontologies extend ordinary finite ontologies with rules for producing recursively shaped terms representing the ontological content (ontological semantics) of NL noun phrases and other phrases. We focus here on achieving a robust, often only partial, ontology-driven parsing of and ascription of semantics to a sentence in the text corpus. The aim of the ontological analysis is primarily to identify paraphrases, thereby achieving a search functionality beyond mere keyword search with synsets. We further envisage use of the generative ontology as a phrase-based rather than word-based browser into text corpora.