Nearly-automated metadata hierarchy creation

  • Authors:
  • Emilia Stoica;Marti A. Hearst

  • Affiliations:
  • University of California, Berkeley, Berkeley CA;University of California, Berkeley, Berkeley CA

  • Venue:
  • HLT-NAACL-Short '04 Proceedings of HLT-NAACL 2004: Short Papers
  • Year:
  • 2004

Quantified Score

Hi-index 0.00

Visualization

Abstract

Currently, information architects create metadata category hierarchies manually. We present a nearly-automated approach for deriving such hierarchies, by converting the lexical hierarchy WordNet into a format that reflects the contents of a target information collection. We use the term "nearly-automated" because an information architect should have to make only small adjustments to produce an acceptable metadata structure. We contrast the results with an algorithm that uses lexical co-occurrence statistics.