Design for IDSSWhat is design?Slide 3Classifying Design TaskRoutine DesignInnovative DesignComplex DesignComplex and Innovative Tasks (1)Complex and Innovative Tasks (2)Case Based DesignSlide 11Slide 12Example of Complex Case UsageSlide 14Slide 15Knowledge RepresentationVocabularyCase BaseCase Base (cont)Similarity MeasureComputational ApproachRepresentational ApproachCBR for Innovative and Creative DesignSolution Transformation and Case AdaptationSlide 25Fish and ShrinkSlide 27Fish and Shrink (2)Structural SimilaritySlide 30Slide 31Slide 32Slide 33Case Study – EADOCSEADOCS (2)EADOCS (3)EADOCS (4)EADOCS (5)EADOCS (6)Final RemarksReferencesDesign for IDSSLiam PageCSE 43523 October 2006What is design?Construction of an artifact from single parts that may be either known and given or newly created for this particular effect (Börner 1998)Design systems assist a user in producing better designs in shorter amount of timeWhat is design?How does design help with:decreasing design times?increasing design quality?improving design predictability?Classifying Design TaskThree classifications:Routine DesignInnovative DesignCreative DesignRoutine DesignState space is well defined using potential designsNew designs can be derived entirely from existing designsOutcomes known before handFinal design agrees with configurable constraintsUsed mostly in KB-systemsInnovative DesignWell defined state space of potential designs, non-routine design desiredValues for variables may changeSolution is similar to old designs, but also appears to be new due to variablesComplex DesignNon-routine designNew variablesExtends/moves state space of potential designsComplex and Innovative Tasks (1)Often unsure what the final design constraints will beTypically ordered in accordance to preference criteriaAbstract -> ConcreteReduction of design spaceComplex and Innovative Tasks (2)Ideal systemAssists user, not automatedUser interface logically constructed for type of design taskLearns from past solutions and user’s response to solutions (accept, correct, refuse)Case Based DesignThemes of case based designed systems (Maher and Gomez de Silva Garza 1997)representation and management of complex casescase augmentation using generalized design knowledgeformalization of informal knowledgeCase Based DesignWhat can be a complex case?Sample of larger data modelData represented structurally (graphs)Non-static variablesFlexible – may have multiple interpretationsAdaptable to solve new problemsCase Based DesignImplications of complex casesMust be able to reinterpret and reformulate new problemsOverlapping of problem and past cases must be identifiedParts must be chosen for transfer and combinationSimilarity functions must be flexibleJoint consideration of case aspects is possibleExample of Complex Case UsageCase: DeluxeBathroom1• Dimensions = (20’-40’)x (20’-40’)• Doors = 1 – 2• Outlets = 4 – 6• Hot tub = yes• …Case: DeluxeBathroom2• Dimensions = ( 30’ 50’)x (30’x50’)• Doors = 2 – 3• Outlets – 6 – 10• Deluxe Standing Shower = yes• …Transformed Solution• Dimensions = 30’ x 30’• Doors = 1• Outlets = 6• Deluxe Standing Shower = yes• …Case Based DesignGeneralized design knowledge to augment casesIncludes causal models, state interactions, heuristic models/rules, geometric constraintsTypically not available for innovative and creative tasksCase Based DesignNeed formalization of knowledge for CBR automationProblem: human knowledge of design is difficult to formalize into rules and variables that the system can utilizeIn cases where it is only possible to create an informal body of knowledge, system should be developed to merely support a human designerKnowledge RepresentationFour knowledge containers in CBRVocabularyCase baseSimilarity measureSolution transformationVocabularyVocabulary – task and domain dependentShould capture all important features of designSupports problem solving in relevant domainCase BaseRepresent past design experienceUsage – abnormal/normalGranularity – grain size of cases is equal to grain size of design taskLevel of AbstractionOssified cases – general rules of thumbParadigmatic cases – represent learned expertise Stories – complex, relate to large number of circumstanceCase Base (cont)PerspectiveState-oriented – case represents problem and solutionSolution-path – case refer to problem or operator that determines solution from problem descriptionSimilarity MeasureTwo different approaches to similarity assessmentComputational (similarity) approachRepresentational approachComputational ApproachUnstructured organizationUsefulness of cases based on presence or absence of featuresMany cases Are Called – candidate casesFew Are Chosen – structural comparison between problem and possible solutionsRepresentational ApproachPre-structured case base (indexing structure)Neighboring cases are assumed to be similarProbes constraints in memory to determine possible solutionsCBR for Innovative and Creative DesignFlexible case retrievalRetrieved cases show similar aspects to the problemDifferent similarity measures have to be dynamically composed during retrievalFish and Shrink AlgorithmStructural similarity assessmentStructural cases are processed and represented as variables taking the role of problem or solution variablesSolution Transformation and Case AdaptationNew situations often different from old solutionsSolutions must be adapted to fit the constraints of the problem using parts from other past solutionsSolution Transformation and Case AdaptationThree kinds of adaptation (Cunningham and Slattery 1993)Parametric adaptation – modifying parametersStructural adaptation – adaptation operators (grammar rules)Generative Adaptation – reuse and adaptation for derivations of past problem-solving episodesFish and ShrinkAlgorithm for flexible case retrievalAllows for rapid searching through case base (even if significant aspects are combined at query time)Can be stopped at any time and still produce usable results (though not complete)Fish and ShrinkSimilarity measure of emphasized attributes between all cases and a set of test cases are retrieved and