The Automated Refinement of a Requirements Domain Theory

  • Authors:
  • T. L. McCluskey;M. M. West

  • Affiliations:
  • School of Computing and Mathematics, The University of Huddersfield, Huddersfield HD1 3DH, UK. t.l.mccluskey@hud.ac.uk;School of Computing and Mathematics, The University of Huddersfield, Huddersfield HD1 3DH, UK. m.m.west@hud.ac.uk

  • Venue:
  • Automated Software Engineering
  • Year:
  • 2001

Quantified Score

Hi-index 0.00

Visualization

Abstract

The specification and management of requirementsiswidelyconsidered to be one of the most important yetmostproblematicactivities in software engineering. Insomeapplications, such asin safety critical areas orknowledge-basedsystems, theconstruction of a requirements domaintheory isregarded as animportant part of this activity. Buildingandmaintaining such adomain theory, however, requires alargeinvestment and a range ofpowerful validation and maintenancetools.The area of itheoryrefinement is concerned with the useoftraining data toautomatically change an existing theory so thatitbetter fits thedata. Theory refinement techniques, however,havenot beenextensively used in applications because of theproblems inscalingup their underlying algorithms.In our work we have applied theory refinement to assistintheproblem of validation and maintenance of arequirementstheoryconcerning separation standards in the North EastAtlantic.In thispaper we describe an implemented refinementalgorithm,whichprocesses a logic program automatically generatedfrom thetheory.We overcame the size and expressivenessproblemstypicallyencountered when applying theory refinement to alogicprogram ofthis kind by designing focused, compositerefinementoperatorswithin the algorithm. These operators modifytheauto-generatedlogic program by generalising or specialisingclausescontainingordinal relationsthat is relations which operateontotally ordereddata.