• DocumentCode
    2202707
  • Title

    Heuristic metrics for AI problem domains such as natural language processing

  • Author

    Facemire, Jon

  • Author_Institution
    Dept. of Comput. Sci., Alabama A&M Univ., Normal, AL, USA
  • fYear
    1989
  • fDate
    26-28 Mar 1989
  • Firstpage
    658
  • Lastpage
    661
  • Abstract
    It is proposed that a series of numeric metrics be established for the evaluation of problem domains such as natural-language processing. These metrics should be weighted averages of multiple independent factors and should be constructed so that typical human-level performance is taken to be unity. The usage of such metrics is discussed, and the example of natural language processing is detailed with a suggested metric delineated
  • Keywords
    artificial intelligence; heuristic programming; natural languages; performance evaluation; AI problem domains; natural language processing; numeric metrics; Artificial intelligence; Computer science; Face; Humans; Measurement standards; Natural language processing; Pattern recognition; Project management; Software packages; Standards development;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    System Theory, 1989. Proceedings., Twenty-First Southeastern Symposium on
  • Conference_Location
    Tallahassee, FL
  • ISSN
    0094-2898
  • Print_ISBN
    0-8186-1933-3
  • Type

    conf

  • DOI
    10.1109/SSST.1989.72549
  • Filename
    72549