Combining structured and unstructured knowledge sources for question answering in watson

  • Authors:
  • Ken Barker

  • Affiliations:
  • IBM Research, Thomas J. Watson Research Center, Yorktown Heights, NY

  • Venue:
  • DILS'12 Proceedings of the 8th international conference on Data Integration in the Life Sciences
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

One of the classical challenges of Artificial Intelligence research has been to build automatic, open-domain question answering (QA) systems. The goal is not merely to retrieve documents containing answers to questions, or to query databases known to contain the answers. Rather, open-domain question answering systems must accept any question on any topic, find relevant information from possibly disparate sources, synthesize an answer, explain the evidence supporting the answer and provide an indication of the systems confidence that the answer is correct.