Adapting a general semantic interpretation approach to biological event extraction

  • Authors:
  • Halil Kilicoglu;Sabine Bergler

  • Affiliations:
  • Concordia University, de Maisonneuve Blvd. West, Montréal, Canada;Concordia University, de Maisonneuve Blvd. West, Montréal, Canada

  • Venue:
  • BioNLP Shared Task '11 Proceedings of the BioNLP Shared Task 2011 Workshop
  • Year:
  • 2011

Quantified Score

Hi-index 0.00

Visualization

Abstract

The second BioNLP Shared Task on Event Extraction (BioNLP-ST'11) follows up the previous shared task competition with a focus on generalization with respect to text types, event types and subject domains. In this spirit, we re-engineered and extended our event extraction system, emphasizing linguistic generalizations and avoiding domain-, event type-or text type-specific optimizations. Similar to our earlier system, syntactic dependencies form the basis of our approach. However, diverging from that system's more pragmatic nature, we more clearly distinguish the shared task concerns from a general semantic composition scheme, that is based on the notion of embedding. We apply our methodology to core bio-event extraction and speculation/negation detection tasks in three main tracks. Our results demonstrate that such a general approach is viable and pinpoint some of its shortcomings.