Ideation Methods Building on Fundamental Studies in Design by ...

NSF GRANT CMMI-0555851NSF PROGRAM NAME: Engineering Design<strong>Ideation</strong> <strong>Methods</strong> <strong>Building</strong> on FundamentalStudies in Design by AnalogyChristina K. WhiteTeachers CollegeUniversity of ColumbiaKristin L. Wood and Jason WeaverManufacturing and Design LabThe University of Texas at AustinAbstract: This project seeks to investigate, develop,test, and implement innovative design methods forgenerating concepts through analogies. Design-byanalogyis the area of idea generation in whichproven solutions from a given domain provide visual,aesthetic, geometric, or functional similarity forproblem-solving in another domain. The existence ofthis area is noted throughout recorded history. Thefirst steam locomotives, for example, were based onanalogies to the horse-drawn stage coach. While theexistence of analogies as inspiration for creativedesign is well known, the cognitive processes used indesign-by-analogy, especially for complex problemssuch as in engineering, is far from understood. Thispaper uses results from our fundamental studies ofdesign-by-analogy and considers a study ofalternative ideation methods. In particular, the designmethods, known as Transformation Design andWordTree Design-by-Analogy, seek to reformulatedesign problems so that multiple effective analogiesmay be developed. A full-factorial analysis was usedto compare the effect of using the two designmethods. Results from the design teams wereevaluated quantitatively by the number conceptsgenerated.1. IntroductionDesign-by-analogy is a powerful tool for developingideas. Designers frequently base their designs onconcepts or systems they have seen before [1-4].Numerous examples of innovative products based onanalogies are included in various forms of mediasuch as technical magazines, literature, even films,and also in nature and other areas that are not asobvious prompts for connectivity. We seek in thispaper to explore concept generation techniques thatsupport Design-by-Analogy.In design, concept generation is a dynamic area thatcontinues to undergo development. Numerousportrayals of the design process describe [5-8]concept generation as a key part of any approach.Over time, concept generation transforms from vagueideas of brainstorming to fleshed-out formal methodssuch as mind-mapping [7], morphological matrices[7], 6-3-5/C-Sketch [7], and TIPS/TRIZ [9].In this project, we pursue research on severaldifferent fronts related to concept generation and thedesign process in general. The first area of researchexamines Design-by-Analogy techniques andformulates a method for concept generation using acommon verb terms found in natural languages. Thesecond area deals specifically with systems that cantransform between different configurations in orderto perform multiple functions. A large number ofanalogous “transformers” are examined, andinformation gleaned from them can be used in thedesign of new systems.2. Background2.1 WordTree Design-by-Analogy. Designers areoften inspired by analogies in the process ofdeveloping new ideas. In fact, a large portion ofinnovation is derived directly from analogy.However, much of this process is ad hoc orserendipitous. The challenge and purpose of a formalDesign-by-Analogy method is to lead the designer touseful, but non-obvious analogies. This guidingprocess usually involves re-representing the problemfrom a different point of view. For example, somemethods present the problem in terms of similarbiological systems [10,11], while others describe itskey functions and flows [12]. Synectics maychallenge the designer “to be” the problem or look atit symbolically [13].WordTree Design-by-Analogy is a recentdevelopment in design methodology [14-16]. In thismethod, key words such as functions or customerneeds are used to seed a linguistic representation ofthe problem. This representation is developed boththrough an extended brainstorming approach andwith the help of WordNet, a tool developed atPrinceton University [17,18]. The WordNet databasefunctions by returning a list of words related to aProceedings of 2009 NSF Engineering Research and Innovation Conference, Honolulu, HawaiiGrant CMMI-0555851

Figure 1: Exemplar WordNet results andWordTree for key word “track” [17].user-supplied seed word. WordNet works similarly toa thesaurus, but gives much more functionality andstructure. For each input word, WordNet returns a setof troponyms (sub-types or more specific synonyms),hypernyms (overarching or more general synonyms),and sister terms (other words stemming from acommon hypernym). An example WordNet entry forthe verb “track” is shown in Fig. 1. The informationretrieved from WordNet can easily be assembled intoa graphical representation called a WordTree, withhypernyms and troponyms branching off of the seedterm like roots and branches. An example WordTreebased off of the same key word “track” is shown inFig. 1.Hence the name, the information in a WordTree canbe used to branch out from known functions andcustomer needs into far-field applications. Forexample, a sample design problem may require thefunction “track,” as in Fig. 1. The associatedWordTree terms may lead the designer to explorecurrent applications of “hunt” (perhaps hunting for asolution to a problem or hunting for a lost item). Itwould also lead to ways to “scout” or “reconnoiter”(methods and skills used by explorersor pioneers),and tools to “monitor” or “trace” (to view thesituation from different vantage points – perhapsthrough electronics). These trains of thought wouldreveal specific, real-world examples that might haveotherwise gone unnoticed. By exploring multipleWordTrees, the designer is led towards a wide swathof related concepts and applications far-removedfrom the initial problem, i.e., design-by-analogy.Figure 2 shows a more extensive and complete formof the WordTree Design-by-Analogy method. Noticethat the method provides scaffolding bysystematically guiding a designer or design teamthrough the exploration of analogous verb terms andanalogous domains. An exemplar design result is alsoshown, providing design concepts for the problem ofassisting persons with disabilities in folding clothesas part of their daily jobs.2.2. Transformation Design. Transformation, in amechanical sense, can be defined as changing state(or configuration) in order to provide newfunctionality [19,20,21]. Transforming, orreconfigurable systems occur frequently in design,ranging from vehicles to furniture to toys and tools.From the comics to toy chests to the movie screen,the principles and facilitators of transformation tapthe imagination of many who may not have initiallybeen intrigued by and been exposed to the elegantengineering concepts. This 20 th century a pop culturephenomenon of Transformers is a scaffolding toolthat potentially connects to teaching and learning 21 stcentury engineering.A basic example of a transformer is shown in Fig. 3.Here, a chair transforms into a stepstool by foldinghalf of the structure about a hinge.Figure 2: Complete WordTree Design-by-AnalogyMethodology and Exemplar Result [16,17].Proceedings of 2009 NSF Engineering Research and Innovation Conference, Honolulu, HawaiiGrant CMMI-0555851

Table 1: Transformation principles and facilitators [20].Expand/CollapsePrinciplesExpose/CoverFuse/DivideConformwithStructuralInterface[23]FlipInterchangeWorking OrganEnclose Furcate ModularizeFacilitators[22]Roll/Wrap/CoilShare CoreStructureShare PowerTransmissionShellFan Inflate Nest Share Functions Telescope[24]UtilizeCompositesUtilizeGenericConnectionsfreedom to pursue unconventional match-ups amongthe principles and facilitators.Figure 3: Transforming chair/stepstool [22].Transformers can be complex and difficult toexecute, yet their historic design has been mostly adhoc, relying on serendipitous analogies and previousexperience to create smooth, functionaltransformation processes. Over the past three years,extensive research has been conducted to betterunderstand how transformation works, how it iscurrently designed, and how a better, moreformalized design process could be applied. Over thecourse of this research, a theory of transformationdesign has been developed which identifies a set ofprinciples and facilitators relevant to the process oftransformation (Table 1). These principles andfacilitators represent meta-analogies to generateconcepts for design problems.Additional research shows that these principles andfacilitators occur in consistent, predictable patterns[20,21]. These trends and interrelationships can beutilized to assemble transformation processes thatoperate smoothly and robustly, while still having the3. Research ExperimentA controlled experiment was designed with the aimof comparing the effect of using WordTrees and/orTransformation Design methods in the conceptgeneration process. To focus on these two methods,other parts of the process were controlled as much aspossible.Design teams of three to six members generatedconcepts for a design problem; specifically to create asystem that can tag and track a vehicle for emergencypurposes, then perform an effect on the target whendesired, in this scenario, to aid the search and rescueteam in events to help with hostage situations. Anillustration of this scenario is shown in Fig. 4. Teammembers followed a script outlining the methods tobe used and recorded all concepts for subsequentanalysis.3.1 Source of Participants. Forty-one participantswere assembled into nine teams of three to six peoplefor this study. All participants were volunteerundergraduate students at The University of Texas atAustin. Approximately 75% of them were male, 25%were female. The volunteers’ ages ranged from 21-27. At the time of the study, they were enrolled in asenior-level mechanical engineering capstone designclass, where they had previously learned commonconcept generation techniques such as mind-mappingProceedings of 2009 NSF Engineering Research and Innovation Conference, Honolulu, HawaiiGrant CMMI-0555851

and 6-3-5. The volunteers had experience typical ofundergraduate engineering students, with somereceiving additional training through internships, coops,previous jobs, and research opportunities.Because of their general engineering experience andcommon level of understanding of design as taught inthe class, it was assumed that they would havecomparable backgrounds coming into the test study.Figure 4: Example scenario involving "tag, track, and effect" design problem.3.2 Factorial Organization. The study wasformulated as a basic 2 2 factorial experiment withfour different conditions. The two controlledvariables were the use of the WordTree Design-by-Analogy method and the use of the TransformationDesign method.In the first trial (- -), neither WordTrees nortransformation heuristics were used. This conditioncorresponds to a control. After a brief introduction, adesign problem was given, after which theparticipants spent 25 minutes brainstorming with amind-map. This was followed by 30 minutes of 6-3-5. In the final stage of the process, each participantsearched the Internet (through patents, images,videos, and design websites) for additionalinformation on analogies uncovered during the earlierstages. A research assistant introduced each step ofthe process, using a script for consistency in languageand timing. The complete outline for this “control”variation is given in the next section.In the second trial (+ -), a WordTree Design-by-Analogy method was inserted before the mindmapping.This method is also outlined in the nextsection. The third trial (- +) inserted an introductionto transformation principles and facilitators, withinformation on how to use them. The fourth trial (++) used both WordTrees and transformationheuristics3.3 Outline of Experiment. Each instance of theexperiment followed the outline shown below. Thecontrol (- -) trial follows the steps as listed, excluding3A and 3B (shown in italics). The other variationsinserted steps 3A (for WordTrees) and/or 3B (forTransformation Design) as appropriate. The assistantleading the session was given a script to read to theparticipants, and each stage was given a set timeperiod, with the total session lasting two hours.1. Introduction to “transformers” – The students aregiven a definition and brief explanation oftransformation, and they are invited to give examplesof transformers to demonstrate understanding.2. Introduction to analogies – The students are givena brief explanation of how analogies are often usedcognitively in design and how actively searching foranalogies can lead to benefits in the design process.Proceedings of 2009 NSF Engineering Research and Innovation Conference, Honolulu, HawaiiGrant CMMI-0555851

form by a different member of the group. If the sameconcept was listed twice in the same stage (e.g. mindmapping),only one contribution was recorded. If adifferent team member significantly developed theconcept at a later stage (such as 6-3-5), thiscontribution was also recorded, but was notconsidered unique. Thus, the data as a whole can beexamined by the total number of concepts generatedor the number of unique concepts generated.Figure 5: Sample 6-3-5 result showing burrowingtagBecause the number of participants varied amongdifferent groups from three to six, it is useful toapproach the aggregate data from several differentangles. We chose to evaluate the experiment usingthree different aspects of the data:1. The total number of concepts generated byeach group (including multiple instances).2. The number of unique concepts generatedby each group.3. The number of concepts generated by eachindividual.In this way, we avoided several issues with the data.If we only look at totals for the groups, the resultsmay be skewed because of differences in group size.By looking at data for both groups and individuals,we can make full use of a larger population size butstill observe group-specific trends.4.2 Statistical Analysis of Data by Group. The firstanalysis of the data examines the total number ofconcepts generated by each group. Of the ninegroups, eight are included here, with two groups ateach data point. The last group is a third case of usingWordTree, and was omitted because it is the onlygroup with 6 people. The results can be organized forDesign-of-Experiments as shown in Table 2.Design-of-Experiments provides a least-squaresestimate for the effect of each factor and interactionin the experiment. In the table above, the numbers ofconcepts generated by each of the eight includedgroups are listed under p1 and p2. The effects of thevariables d1 (WordTrees) and d2 (TransformationDesign), as well as the interaction between them aredetermined by subtracting the average for all “-1”runs from the average for all “+1” runs. We see that,on average, using WordTrees tended to produce 10less concepts than not using it, while usingtransformation principles and facilitators led to 29more concepts than not using them. The interactionbetween the two methods gave an effect of 7 moreideas.However, it is important to determine how much ofthis information is due to actual trends and how muchis due to error and random variation. Because wehave two replicates at each data point, we cancompute the variance for each run. From this, we canfind the pooled variance (S p 2 = average of variances)and the standard deviation of the effects (S e = 2S p /√mwhere m = number of groups). A t-value for eachfactor effect is also determined by dividing theestimated effect by S e . These t-values can then becompared to required t-values for four runs andconfidence levels of 95% and 99%.As shown in Table 3, the use of TransformationDesign (d 2 ) has a t-value of 4.7, which indicates witha 99% confidence level that the effect of usingTransformation Design is statistically significant. Theuse of WordTrees did not prove to be significant inthis instance, nor did the interaction between the twomethods. ANOVA calculations confirm the relativesignificance of the effects.Table 3 shows the results of the second analysis ofthe data. Here, only the number of unique concepts isincluded for each group; instances where a concept isdeveloped further later in the process are notrecounted. This analysis shows that once again, theuse of Transformation Design (d 2 ) was the onlysignificant factor, with a confidence level of 99%.Proceedings of 2009 NSF Engineering Research and Innovation Conference, Honolulu, HawaiiGrant CMMI-0555851

Table 2: Factorial results using total number of concepts per groupTrial Mean d 1 (WordTree) d 2 (Transformation) d 1 d 2 p 1 p 2 Average Variance1 +1 -1 -1 +1 77 93 85 1282 +1 +1 -1 -1 60 76 68 1283 +1 -1 +1 -1 108 106 107 24 +1 +1 +1 +1 109 99 104 50Σ⋅Π: 364 -20 58 14 S p : 8.77Effect: 91 -10 29 7 S e : 6.2t E : 14.7 -1.6 4.7 1.1 t* 95 (4): 2.78t* 99 (4): 4.60Table 3: Factorial results using total unique concepts per groupTrial Mean d 1 (WordTree) d 2 (Transformation) d 1 d 2 p 1 p 2 Average Variance1 +1 -1 -1 +1 75 86 80.5 60.52 +1 +1 -1 -1 56 74 65 1623 +1 -1 +1 -1 95 102 98.5 24.54 +1 +1 +1 +1 101 98 99.5 4.5Σ⋅Π: 343.5 -14.5 52.5 16.5 S p : 7.9Effect: 85.9 -7.25 26.3 8.3 S e : 5.6t E : 15.3 -1.3 4.7 1.5 t* 95 (4): 2.78t* 99 (4): 4.60Table 4: Factorial results using number of concepts per individual, blocked by groupTrialMeand 1d 2(Trans.)d 3(Block)d 1 d 2 d 1 d 3 d 2 d 3p 1 p 2 p 3 p 4 p 5 Avg. Var.(WT)1 +1 -1 -1 -1 +1 +1 +1 16 25 15 21 19.3 21.62 +1 +1 -1 -1 -1 +1 -1 14 16 18 12 15 6.73 +1 -1 +1 -1 -1 -1 +1 28 34 20 26 27 33.34 +1 +1 +1 -1 +1 -1 -1 41 32 36 36.3 20.35 +1 -1 -1 +1 +1 -1 -1 17 15 20 15 27 18.8 25.26 +1 +1 -1 +1 -1 -1 +1 19 15 13 17 12 15.2 8.27 +1 -1 +1 +1 -1 +1 -1 32 22 26 14 12 21.2 69.28 +1 +1 +1 +1 +1 +1 +1 15 19 21 21 23 19.8 9.2Σ⋅Π: 172.6 0.1 36.1 -22.6 15.8 -22.1 -10.1 S p : 4.9Effect 21.6 0.0 9.0 -5.7 4.0 -5.5 -2.5 S e : 1.7t E : 12.8 0.0 5.4 -3.4 2.3 -3.3 -1.5 t* 95 2.31t* 99 3.364.3 Statistical Analysis of Data by Individual. Onechallenge of interpreting this study by group is therelatively small sample size. With only two replicatesfor each of the four trials, uncertainty exists about thetrue characteristics of the population, even though theresults are statistically significant. Different groupsizes, personalities and interactions in the groups, andeven time of day may affect the studied factors. Wecan increase the effective sample size if we alsoexamine the results by individual. Now, instead ofeight (or nine) samples, there are 35 (or 41)independent samples. The additional unknowns thatresult from different group dynamics can be controlledby using blocks to separate the individuals of differentgroups. Hence, the results can be arranged as shown inTable 4 (once again leaving out the 6-person group toeven out the distribution of replicates throughout theexperiment).Like the group results, this analysis indicates that usingthe transformation principles and facilitators has aProceedings of 2009 NSF Engineering Research and Innovation Conference, Honolulu, HawaiiGrant CMMI-0555851

statistically significant positive effect on the designprocess (confidence level of 99%). The greater samplesize also results in eliminating most of the effect due tothe use of WordTrees. In addition, the effect of theblock (members of different groups) was also found tobe significant (confidence level of 99%).Another insight resulting from this analysis ofindividual contributions is that there is a significantinteraction between the use of WordTrees andTransformation Design. Even though WordTrees bythemselves did not contribute to a rise in quantity, theinteraction between the two design methods resulted ina slight increase in the number of concepts, above andbeyond the effect of Transformation Design alone(confidence level of 95%).necessity to (re)consider innovation, imagination, andliteracy as ways to develop and strengthen learningthat beacons engineering education to value andaddress the dynamic needs of design. Fromrepresentatives of the United Nations, to policymakers,educators, economists, artists, engineers, tomany in between, the importance of world citizens tobe critical thinkers, innovators, and have literacy skillsfor the 21 st century is lauded. Indeed, the Director-General of the United Nation Education, Science,Community Organization posits, “In a worldincreasingly shaped by science and technology,scientific and technological literacy is a universalrequirement… it is vital to improve scientific andtechnological literacy” [29].4.4 Diversity and Novelty of Solutions. The dataindicate a large degree of variation between teams inthe concepts that were generated. Almost 500 differentconcepts were generated between the nine groups, withan average of 90 concepts per team over a two-hourperiod. Out of these, 352 concepts were onlymentioned by one team. In fact, although a fewconcepts were proposed as many as 12 times, theaverage number of times any given concept wasgenerated over the course of the experiment was only1.66, showing a wide spread of diversity. Thisdiversity was supported by the WordTree method eventhough the experimental results do not show anincrease in quantity of ideas. Thus, the WordTreemethod does assist in increasing novelty and diversity.The concepts generated seemed to fall into sixcategories. The majority of them fulfilled the mainfunctions of “tag, track, and effect” from the designstatement. In addition, some concepts addressed theoverall form of the system, such as resembling a bird,insect, seedpod, helicopter, or animal droppings. Otherconcepts developed general methods of power ortransportation (e.g. hovering, insect legs, diggingunderground), as well as transformation issues (foldingwings, changing shape for camouflage, break-awaymodules for tagging, etc.). An example concept oflooking to nature of analogous design is shown in Fig.6, where a bird-like UAV releases a tag embedded in asubstance disguised as bird droppings.5. Discussion: Broader ImpactsThe results from the previous section inform not onlyinnovators in engineering but also engineeringeducation. It is vital for engineering educators to haveresources that integrate powerful tools of conceptgeneration which resonate with complex and differentways of thinking as classrooms are increasingly filledwith diverse learners. There is an internationalFigure 6: Exemplar concept from 6-3-5.Whereas traditional engineering education pedagogyresonates with notions of transmitting knowledge fromone person (usually the professor or engineer) toanother (usually the student), DbA, WordTree DbA,Transformation design, 6-3-5, and mindmaps integratepedagogy gleaned from education and psychologyresearch that helps illuminate ways that designersthink, generate ideas, learn, and develop literacy inscience, technology, engineering, and mathematics(STEM.) This research on the development andinclusion of methods that build on psychological andeducational tools to improve innovation techniques anddevelopment of literacy in engineering, integratesmultiple modes of communication. With thesescaffolding tools in DbA, designers are more able todescribe and investigate complex engineering conceptsand generate innovative ideas, thus STEM becomesaccessible to a much broader range of communitieshaving varied degrees of exposure to engineering.Proceedings of 2009 NSF Engineering Research and Innovation Conference, Honolulu, HawaiiGrant CMMI-0555851

Breaking down the communication barrier then opensthe space for increased novel idea generation.In this research formal definitions of terms provide astandard language by which people can discusstransformers, thereby enhancing designers’understanding of transformation processes similar tothe language found in the field of kinematics androbotics. The nature of language, engineering languagein this context, is heteroglossic – meaning that as moreand different people engage in dialogue aboutengineering, the language shifts, changes, andencompasses new meanings over time [30]. Applyingthe principles and facilitators imaginatively to identifysub-systems of transformation, integrate these subsystems,develop the kinematic skeleton, and finallygenerate the storyboarding of transformation producesmultiple and varied language acquisition experiencesthus increasing engineering literacy. The reading,speaking, and producing of multiple texts (spokenlanguage, images, products, symbols) with thescaffolding of principles and facilitators definitionssupports the participants in the heteroglossicdiscussions of engineering design and concepts.Analogies of design relate across content and contexts.When applying strategies to generate multiple novelideas, techniques that meet the learning styles andvaried methods of communication are more fruitful.Diverse design teams come to the ideation processwith many and different lived experiences. Thesevaried experiences inform the way that each designerviews and questions the world. This world, designedby engineering, is abundant with opportunities forcommunity development through a morecomprehensive understanding about effectivecommunication of idea sharing and problem solving.At times, the hindrance in community development is alack of communication strategies, especially about ideasharing and problem-solving. Communication barriersare community barriers by creating silences andmarginalization of those not able to understand and/orbe actively involved in the engineering design process.The guiding tools in this research to improve conceptgeneration in engineering are methods of increasingways that students in engineering education anddesigners are able to generate and communicate theirideas that ultimately help shape our world.6. ConclusionsThis paper focuses on the study of Design-by-Analogyconcept generation techniques that build uponfundamental studies in DbA, such as the need toreformulate design problems to increase the numberand diversity of analogies developed. A controlledexperiment was conducted using a sample of 41senior-level mechanical engineering students dividedover nine teams. These teams each followed one offour formal concept generation approaches,incorporating two recently developed DbA designmethods: WordTree Design-by-Analogy andTransformation Design. The four conditions testedwere a conventional concept generation process withmind-mapping and 6-3-5, Transformation Designcombined with mind-mapping and 6-3-5, WordTreeswith mind-mapping and 6-3-5, and bothTransformation Design and WordTrees with mindmappingand 6-3-5. The average number of conceptsgenerated in each 2-hour trial was 91 per group, or 22per person. Design-of-Experiments analysis revealedthat while using the WordTree method had nonoticeable effect on the number of concepts generated,using the Transformation Design method yielded aconsistent increase of 25-30% more concepts across allgroups and individuals. Both WordTree andTransformation Design contributed to increaseddiversity and novelty across the design teams. Theseprovide positive support for using advanced ideationtechniques in Design-by-Analogy and have strongimplications on education as broader impacts of thiswork.7. Acknowledgements: The authors would like toacknowledge the support provided from NationalScience Foundation under Grant No. CMMI-0555851and the Cullen Endowed Professorship in Engineering,The University of Texas at Austin.References[1] Christensen, B. T., and Schunn, C. D. “Therelationship of analogical distance to analogicalfunction and pre-inventive structure: The case ofengineering design.” Memory & Cognition, (in press).[2] Leclercq, P. and Heylighen, A. “5,8 Analogies perHour,” Artificial Intelligence in Design '02. J. S. Gero(ed.), 285-303, 2002[3] Casakin, H., and Goldschmidt, G., 1999, “Expertiseand the Use of Visual Analogy: Implications forDesign Education,” Design Studies, 20(2), pp. 153-175.[4] Linsey, J., Laux, J., Clauss, E. F., Wood, K., andMarkman, A., 2007, “Increasing Innovation: A Trilogyof Experiments towards a Design-By-AnalogyMethod,” Proceedings of ASME Design Theory andMethodology Conference, Las Vegas, NV.[5] Ullman, D., The Mechanical Design Process,McGraw Hill, 1997, NY.[6] Ulrich, K., Eppinger, S., 2000, Product design andDevelopment, McGraw Hill, NY.[7] Otto, K., Wood, K., 2001, Product Design:Techniques in Reverse Engineering and New ProductDevelopment, Prentice Hall, NJ.Proceedings of 2009 NSF Engineering Research and Innovation Conference, Honolulu, HawaiiGrant CMMI-0555851

[8] Dym, C., 2000, Engineering Design: A ProductBased Introduction, Wiley, NJ.[9] Altshuller, G. S., 1984, Creativity as an ExactScience: The Theory of the Solution of InventiveProblems, Gordon and Breach, Brussels.[10] French, M., 1988, Invention and Evolution:Design in Nature and Engineering, CambridgeUniversity Press, Cambridge, UK.[11] Hacco, E., Shu, L. H., 2002, “BiomimeticConcept Generation Applied to Design forRemanufacture,” Proceedings of the DETC 2002,ASME 2002 Design Engineering TechnicalConferences and Computer and Information inEngineering Conference, Montreal, Quebec, Canada.[12] McAdams, D., Wood, K., 2002, “A QuantitativeSimilarity Metric for Design by Analogy,” ASMEJournal of Mechanical Design, 124(2), 173-182.[13] Gordon, W. J. J., 1961, Synectics: TheDevelopment of Creative Capacity, Harper andBrothers, NY.[14] Linsey, J.S., 2007, “Design-by-Analogy andRepresentation in Innovative Engineering ConceptGeneration,” Mechanical Engineering, Austin, TX:The University of Texas at Austin.[15] Linsey, J., Wood, K., Markman, A., 2008,“Increasing Innovation: Presentation and Evaluation ofthe Wordtree Design-by-Analogy Method,” ASMEDesign Theory and Methodology Conference, NewYork, NY.[16] Linsey, J., Wood, K., Markman, A., 2008,“WordTrees: A Method for Design-by-Analogy,”Proceedings of the 2008 ASEE Annual Conference andExhibition, Pittsburgh, PA.[17] “WordNet: a lexical database for the Englishlanguage,”http://wordnetweb.princeton.edu/perl/webwn.[18] Fellbaum, C., 1998, Wordnet, an ElectronicLexical Database, The MIT Press, MA.[19]. Singh, V., Skiles, S. M., Krager, J. E., Wood, K.L., Jensen, D., Szmerekovsky, A., 2006, “Innovationsin Design Through Transformation: A FundamentalStudy of Transformation Principles,” DETC-2006-99575, International Design Engineering TechnicalConferences, Philadelphia, PA.[20] Weaver, J., Wood, K., Jensen, D., 2008,“Transformation Facilitators: A Quantitative Analysisof Reconfigurable Products and Their Characteristics,”DETC2008-49891, International Design EngineeringTechnical Conferences, Brooklyn, NY.[21] Weaver, J., 2007, “Transformer Design: EmpiricalStudies of Transformation Principles, Facilitators, andFunctions,” Mechanical Engineering, Austin, TX: TheUniversity of Texas at Austin.[22] Mollerup P., 2001, Collapsible: The Genius ofSpace Saving, Chronicle Books, CA.[23] Brute B-2 Folding Chair, 2005,http://www.buyersbox.com/deluxe_folding_chairs.html.[24] Black & Decker model KC2002FK, 2008,http://www.lenehans.ie/lenehans/Images/DB_Detail/kc2002fk.jpg.[25] Lawson, J., Erjavec, J., 2001, Modern Statisticsfor Engineering and Quality Improvement, Duxbury,CA.[26] Shah, J. J., Vargas-Hernández, N., and Smith, S.M., 2003,“Metrics for Measuring <strong>Ideation</strong> Effectiveness,”DesignStudies, 24, pp. 111-134.[27] Shah, J. J., Kulkarni, S. V. and Vargas-Hernández, N., 2000, “Evaluation of Idea Generation<strong>Methods</strong> for Conceptual Design: Effectiveness Metricsand Design of Experiments,” Transactions of theASME Journal of Mechanical Design, 122, pp. 377-384.[28] Bouchard, T. J. Jr., 1969, “Personality, Problem-Solving Procedure, and Performance in SmallGroups,” Journal of Applied Psychology Monograph,53(1), pp. 1-29.[29] Mayor, F. Director-General of United NationsEducation, Science, Cultural Organization, 1999,Science and technology in Africa: A commitment forthe 21 st century,http://www.unesco.org/bpi/scitech/print.htm.[30] Bakhtin, M., 1981, The dialogic imagination.Austin, TX: The University of Texas Press.Proceedings of 2009 NSF Engineering Research and Innovation Conference, Honolulu, HawaiiGrant CMMI-0555851