Exaggeration

  • Authors:

  • Daniel S. Weld

  • Affiliations:

  • Univ. of Washington, Seattle

  • Venue:
  • Artificial Intelligence
  • Year:
  • 1990

Quantified Score

Hi-index 0.00

Visualization

Abstract

Exaggeration is a technique for solving comparative analysisproblems by considering extreme perturbations to a system. For example,exaggeration answers the question, “What happens to the outputtemperature of a heat exchanger if fluid flow rate increases?” bysimulating the behavior of an exchanger with infinite flow rate.Exaggeration is implemented as a sequence of three phases: transform,simulate, and scale. The transform phase takes a comparative analysisproblem and generates the description of an exaggerated system. Thesimulate phase predicts the behavior of the transformed system. Finally,the scale phase compares the original and exaggerated behaviors toanswer the original comparative analysis question. This paper explainsthe theoretical basis of exaggeration, describes an implementation thathas solved over fifty problems, and compares exaggeration with thedifferential qualitative (DQ) analysis approach tocomparative analysis.—Author's Abstract