Medical expertext as regularity in semantic nets

  • Authors:
  • Hafedh Mili;Roy Rada

  • Affiliations:
  • Department of Mathematics and Computer Science, University of Quebec at Montreal, P.O. Box 8888 A, Montreal PQ H3C 3P8, Canada;Department of Computer Science, University of Liverpool, Liverpool L69 3BX, UK

  • Venue:
  • Artificial Intelligence in Medicine
  • Year:
  • 1990

Quantified Score

Hi-index 0.00

Visualization

Abstract

Hypertext emphasizes the logical structure of text. If the text is a single document, its outline gives the first detailed impression of this logical structure, while if the text is many documents, a thesaurus commonly gives the first indication of the logical structure of the document collection. Outlines and thesauri help people deal with the complexity of medical text. Our studies explore the patterns which outlines or thesauri should reflect. By exploiting these patterns we add an expert system dimension to hypertext systems and create expertext systems. Patterns in the outlines of two textbooks are considered and then attention is shifted to the thesaurus for the MEDLINE information service. The MeSH thesaurus for MEDLINE is a massive hierarchical knowledge base from which we have extracted parts and added non-hierarchical relations. On two subsets of MeSH, we show ways to add concepts to the thesaurus by following patterns. We establish that a particular pattern of inheritance, called analogical inheritance or regularity, is salient in MeSH.