Identifying candidates for design-by-analogy Paul-Armand Verhaegen a, *, Joris D’hondt a , Dennis Vandevenne a , Simon Dewulf b , Joost R. Duflou a a Mechanical Engineering Department, Katholieke Universiteit Leuven, Celestijnenlaan 300A Bus 2422, Heverlee, Leuven 3010, Belgium b CREAX N.V., Maarschalk Plumerlaan 113, Ieper 8900, Belgium 1. Introduction Design-by-analogy is an ideation or problem solving method based on analogies between products. An analogy is a specific type of similarity, and similarities can be categorized into three categories: attributional similarity, relational similarity, and literal similarity [1]. Fig. 1 depicts this categorization. Attributional similarity, also called surface similarity [2] or semantic relatedness [3], expresses the amount of similarity in attributes between words. Attributes are elements which provide descriptive information [4], e.g. the pen is sharp, or the color of the pen is red. Two types of attributional similarity are distinguished. The first type is semantically associated [5] and encompasses words which are related by frequent association, e.g. pencil and paper. The second type is semantically similar, or taxonomically similar [6] and includes associated words which are more similar to each other, e.g. pencil and pen are both tools to write with. The relational similarity specifies the amount of similarity in the relations between the individual parts comprising the two concepts [4]. It is this type of similarity that is called an analogy, e.g. a river stands to water as a road stands to vehicles. An analogy therefore does not necessitate attributional similarity, e.g. water is not attributional similar to vehicles. However, an analogy which also displays a large amount of attributional similarity among the individual objects of the two concepts is called a literal similarity, e.g. hose stands to water as a pipe stands to water [7]. Hose and pipe both have similar attributes, e.g. cylindrical, and both have a similar relationship to water. In the domain of problem solving and creative idea generation for design, we can interpret a similarity as the link between two products based on certain shared attributes of the products, shared relations between individual parts of the products, or shared functionality [8]. Design-by-analogy is a methodology based on these similarities between products, and features two distinct classes [9]. On the one hand, products designed by extracting solutions from other products within the domain, called within domain design-by- analogy. An example is the designing a lamp foot by extracting ideas and knowledge from other lamp foots. On the other hand, products designed by extracting solutions from other domains, called cross- domain or between domain design-by-analogy. Examples include designing a lamp foot by extracting ideas and knowledge from cranes, mechanical excavators, or animal feet. Fig. 1 depicts the taxonomy of similarity mapped onto the structure of within and cross-domain design-by-analogy. Contrary to what the name makes belief, design-by-analogy encompasses more than only relational similarity (analogy). Within domain design-by-analogy also implies some level of attributional similarity. Only cross-domain design-by-analogy is characterized by little or no attributional similarity and is therefore mapped to the relational similarity (analogy). All other types of analogy involve a higher amount of attributional similarity and are considered within domain design-by-analogy. The term design- by-analogy could therefore be considered misleading and a more correct term would be design-by-similarity. However, due to the frequent use of the term design-by-analogy, this research will follow this term to indicate both within and cross-domain similarity based design or idea generation. Computers in Industry 62 (2011) 446–459 ARTICLE INFO Article history: Available online 11 January 2011 Keywords: Design methodology Design-by-analogy Patent mining Product aspect Design-by-similarity Analogy ABSTRACT Although supported by extensive anecdotal evidence, only recently design-by-analogy has been proven to occur often in problem solving and idea generation. However, the circumstances which facilitate problem solving and creative idea generation by analogy are not well understood and most analogies are not developed by applying a formal design-by-analogy methodology. Furthermore, most software tools which aid in finding and/or explaining analogies are based on manually assembled databases, which require a large amount of interactive work to be constructed and maintained. This paper examines the use of automatically distilled product characteristics, called Product Aspects, as a way to automatically and systematically identify candidate products for design-by-analogy. Case studies illustrate this idea generation methodology for three different target products. ß 2011 Elsevier B.V. All rights reserved. * Corresponding author. E-mail address: paularmand.verhaegen@cib.kuleuven.be (P.-A. Verhaegen). Contents lists available at ScienceDirect Computers in Industry journal homepage: www.elsevier.com/locate/compind 0166-3615/$ – see front matter ß 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.compind.2010.12.007