1. INTRODUCTION
Analogical thinking involves the transfer of information from one domain (the source) to another domain (the target). This analogical transfer is useful when there is some similarity between the source and the target domains (or the relations in those domains) and where that similarity permits reasoning across domains (e.g., Gentner, Reference Gentner, Vosniadou and Ortony1989; Vosniadou & Ortony, Reference Vosniadou, Ortony, Vosniadou and Ortony1989). Where the source domain is familiar and accessible, drawing analogies can make new subjects easier to understand, facilitating the discovery, development, evaluation, and exposition of (natural and social) scientific knowledge (Holyoak & Thagard, Reference Holyoak and Thagard1995, pp. 191, 209). Consequently, analogies are prominently used in many professional practices, including science (Oppenheimer, Reference Oppenheimer1956), medicine (Clarke, Reference Clarke1978), management (Bingham & Kahl, Reference Bingham and Kahl2013), and education (Dupin & Johsua, Reference Dupin and Johsua1989). Analogical thinking is also central to much design activity, where it serves in identifying and solving design problems and in explaining design concepts to others (Christensen & Schunn, Reference Christensen and Schunn2007). Collectively, these aspects of analogical thinking provide the opportunity to generate creative design proposals that lead to innovative products, systems, and services (Dahl & Moreau, Reference Dahl and Moreau2002; Hey et al., Reference Hey, Linsey, Agogino and Wood2008; Kalogerakis et al., Reference Kalogerakis, Lüthje and Herstatt2010; Chan et al., Reference Chan, Fu, Schunn, Cagan, Wood and Kotovsky2011).
One of the most difficult challenges in constructing analogies is the retrieval of a plausible source, especially where the search space is large and where the relationship to the target is not obvious (Holland, Reference Holland1986, pp. 288–289, 312). Such challenges have led to suggestions that it is helpful to have a catalog of possible sources to draw from and some means of identifying those sources that are related to the targets that are being considered (Linsey et al., Reference Linsey, Wood and Markman2008). In response to this need, design researchers have developed computer support tools that assist in the construction and application of both cross-domain analogies (e.g., Chakrabarti et al., Reference Chakrabarti, Sarkar, Leelavathamma and Nataraju2005; Shu, Reference Shu2010; Vattam & Goel, Reference Vattam and Goel2011; Cheong & Shu, Reference Cheong and Shu2012; Goel et al., Reference Goel, Vattam, Wiltgen and Helms2012; Deldin & Schuknecht, Reference Deldin, Schuknecht, Goel, McAdams and Stone2014) and within-domain analogies (e.g., Barber et al., Reference Barber, Bhatta, Goel, Jacobson, Pearce, Penberthy, Shankar, Simpson, Stroulia and Gero1992; Pearce et al., Reference Pearce, Goel, Kolodner, Zimring, Sentosa and Billington1992; Maher et al., Reference Maher, Balachandran and Zhang1995). These tools are often developed to serve two interconnected purposes. First, they represent the types of systems that could be developed to promote and support analogical thinking in design practice. Second, use of these tools provides a basis upon which analogical thinking activities can be studied in observational or experimental research. As such, the tools are both the manifestation of knowledge about analogy-driven design and one of the means by which that knowledge is generated. This paper is intended to support these tool development activities and tool use studies by defining and relating the user requirements for such tools. This is founded on two assumptions. First, developing design support tools according to user-centered requirements promises to decrease development time (by providing clearer goals and information about trade-offs) and also to increase uptake (by improving effectiveness and ease of use). Second, analogy-driven design activities are most easily studied when the tools used to support those activities fit well with the requirements of their users.
The need for better software tools is supported by observations of analogy use in professional design practice. Designers have criticized existing tools for not providing effective mechanisms for identifying and applying knowledge from other domains (Kalogerakis et al., Reference Kalogerakis, Lüthje and Herstatt2010, p. 433). The need to adopt a user-centered approach to analogical design tools is recognized in the research community also, with calls to focus on issues of usability, interface design, visualization, and search (Goel, McAdams, & Stone, Reference Goel, McAdams and Stone2014). This is part of a more general ambition to develop tools that assist people's creative work without disturbing the natural flow of their activities (Hewett, Reference Hewett2005). In defining the user requirements for analogical design support tools, we review three distinct but related areas of literature: the literature on analogical design, creativity support tools, and human–computer interaction (HCI). Analogical design tools are our main focus, but we view them as a subset of creativity support tools, which are thus relevant to how analogical design tools should be developed. More generally, HCI guidelines are also relevant because there are many standards (e.g., ISO 9241 series) and commonly applied heuristics and usability guidelines that should be considered when developing a user interface (e.g., Nielsen, Reference Nielsen, Nielsen and Mack1994; Schneiderman & Plaisant, Reference Schneiderman and Plaisant2005; Galitz, Reference Galitz2007). While developing tools for supporting analogical design, the importance of these generic guidelines is clearly evident. Although emphasizing their importance, this paper does not focus on the generic HCI guidelines, but instead concentrates on the requirements for supporting analogical design activity within the context of creativity support tools.
To identify user requirements relevant to analogical design support tools, we searched for literature sources that describe the development or use of such tools, especially in the context of biomimetics. From the variety of possible analogical sources from which designers might draw, we focus on the biological domain (and biomimetic design tools) for three reasons: because of biomimetics' recent rapid expansion and wide applicability across a range of technologies, problem types, and design disciplines (Lepora et al., Reference Lepora, Verschure and Prescott2013); because biologically inspired design is by definition based on cross-domain analogies and thus promotes or permits a wide range of analogical distances (Goel, Vattam, et al., Reference Goel, Vattam, Wiltgen, Helms, Goel, McAdams and Stone2014); and because there has been much recent tool development work in this area with associated studies of designer interactions with those tools. We also searched for sources that focus on the design requirements for creativity support software more generally. Within the broader scope of HCI research, we searched for requirements relating to how information should be communicated to system users and what interaction qualities support learning and exploration. This review process resulted in the identification of user requirements that fall into two general groups. The first group contains requirements related to the information content that the tool should provide, such as abstraction (generalization from specific instances); the second group of requirements contains interaction qualities that the tool should provide so as to support designers' tasks with the tool, such as accessibility (the ease of accessing the intended content). Developing tools to satisfy these requirements realizes positive outcomes for the analogical design process, either directly or indirectly. To discuss how future tools should be designed, the paper introduces these outcomes and requirements, and examines the relationships between them.
Before proceeding, some clarification of terminology is warranted. We use the term tool to refer to information tools (typically software tools) that assist in constructing analogies that assist with design problems. Designers are the typical users of these tools, and we thus use the terms “designer” and “user” synonymously, switching between these terms depending on the role that we are emphasizing (i.e., the designer as a tool user or the designer as a designer). Where we refer to those who develop these tools, we use the term developer rather than designer to avoid any confusion with the tools' users. [We acknowledge that the tool developers are also designers and that they too could use (analogical) design tools in their work, but we do not consider that matter any further here.]
2. POSITIVE OUTCOMES OF TOOL USE
The ultimate aim of analogical design support tools is to provide stimuli that assist in the construction of analogies and thus facilitate creative design work. This goal can be achieved by the support tools encouraging two different kinds of positive outcome. There are outcomes that directly relate to the construction and implementation of analogies, and there are outcomes that make uptake of the tool more likely and effective and thus indirectly promote analogical thinking. We first consider the direct outcomes and then the indirect ones.
For creative design, many process models have been suggested in the literature, and from those the most commonly identified creativity stages are exploration (or analysis), generation, evaluation, and communication (for a recent review, see Howard et al., Reference Howard, Culley and Dekoninck2008). As illustrated in Figure 1, these stages of creative activity partly map to and partly overlap with the different stages of analogical thinking that have been identified: accessing the analogs from some source (e.g., from memory or from a catalog of some sort); mapping the source analog with the target analog by identifying the correspondence between the two; drawing inferences about the target based on these mappings; evaluating those inferences, considering the requirements of the target, and adapting them based on these evaluations; and finally, learning from the preceding stages to construct new categories and schemas that can be stored as new analogs in long-term memory (Holyoak et al., Reference Holyoak, Gentner, Kokinov, Gentner, Holyoak and Kokinov2001). Of course, neither creative acts generally nor analogical thinking in particular are always carried out in this strict sequence, but the named stages provide a useful basis for structuring how we consider the impacts that design support tools should have on analogical thinking. Furthermore, apart from supporting analogical design stages, tools are expected to assist with problem reframing activities during the design process (Goel et al., Reference Goel, Vattam, Wiltgen and Helms2012).
Fig. 1. Mapping between the stages of analogical thinking and the stages of creative activity (corresponding stages are indicated with two-way arrows).
Besides seeking to directly impact the analogical design process, support tools can indirectly impact that process by creating desirable conditions under which the tool will be used effectively. Having an enjoyable, fun, or playful software environment motivates users to explore the content and spend more time on using the tool to look for ideas (Elam & Mead, Reference Elam and Mead1990; Resnick et al., Reference Resnick, Myers, Nakakoji, Shneiderman, Pausch, Selker, Eisenberg, Shneiderman, Fischer and Hewett2005). Maintaining designer efficiency while working with the tool is also essential, requiring an interactive environment free from interruptions to creative flow. If such efficiency is maintained, users can allocate their resources to design tasks rather than focusing on operating the software (Avital & Te'eni, Reference Avital and Te'eni2009). Finally, it is important to maintain the users' perceived control over the tasks that are carried out with the tool and to decrease the perceived risk of trying something new (Terry & Mynatt, Reference Terry and Mynatt2002; Resnick et al., Reference Resnick, Myers, Nakakoji, Shneiderman, Pausch, Selker, Eisenberg, Shneiderman, Fischer and Hewett2005). When such confidence is combined with intrinsic motivation and a sense of efficiency, the tool can not only help in providing valuable information for creative design but also contribute to establishing a psychological state in which such creative acts are more readily performed (Csikszentmihalyi, Reference Csikszentmihalyi1996).
3. INFORMATION CONTENT PROVIDED BY THE TOOL
Analogical design support tools primarily provide information about the source (e.g., about a biological system). In providing source information, tool developers must make decisions about what entities to describe, how to group them, what representations to use, how to abstract from examples, how to exemplify these abstractions, and so on. These are all questions about information content, and they relate to a critical issue of analogies, that is, how to reveal the salient features of the source while de-emphasizing the irrelevant or distracting features (Halasz & Moran, Reference Halasz and Moran1982; Spiro et al., Reference Spiro, Feltovich, Coulson, Anderson, Vosniadou and Ortony1989).
3.1. Abstraction: Deriving general principles from specific instances
Presenting source information to designers is intended to promote creative thought, but there is the risk that rather than providing inspiring new ways to view a problem, the information instead leads to a blind repetition of unhelpful or limiting features of the stimuli (Jansson & Smith, Reference Jansson and Smith1991; Cardoso & Badke-Schaub, Reference Cardoso and Badke-Schaub2011; Goldschmidt, Reference Goldschmidt2011). This fixation effect has been observed in biologically inspired design activities, where the biological stimuli limit the exploration to that one source (Mak & Shu, Reference Mak and Shu2008; Helms et al., Reference Helms, Vattam and Goel2009). Attempts to remedy this problem can involve one of two related options: either the stimuli are presented in a more abstract form or the stimuli are related in a way that emphasizes general principles rather than detailed features. In either case, some principle of abstraction is used to modify or group the stimuli. This abstraction is often achieved by describing biological entities with function-based ontologies (Chakrabarti et al., Reference Chakrabarti, Sarkar, Leelavathamma and Nataraju2005; Vincent et al., Reference Vincent, Bogatyreva, Bogatyrev, Bowyer and Pahl2006; Nagel et al., Reference Nagel, Nagel, Stone and McAdams2010; Shu, Reference Shu2010; Vattam & Goel, Reference Vattam and Goel2011; Cheong & Shu, Reference Cheong and Shu2012; Goel et al., Reference Goel, Vattam, Wiltgen and Helms2012; Deldin & Schuknecht, Reference Deldin, Schuknecht, Goel, McAdams and Stone2014).
3.2. Exemplification: Illustrating general principles with specific instances
While abstraction is useful for identifying the general principles that apply across domains, concrete exemplifications of these principles are beneficial for the designer to understand the ways in which they are implemented. Especially for novice designers, reasoning from individual cases may be easier than reasoning from abstract principles (Bonnardel, Reference Bonnardel2000; Ball et al., Reference Ball, Ormerod and Morley2004), and the examples may help in transferring those principles into the design (Hewett & DePaul, Reference Hewett, DePaul, Houstis, Rice, Gallopoulos and Bramley2000; Töre Yargın, Reference Töre Yargın2013). For instance, in the AskNature database (Biomimicry 3.8 Institute, 2008–2014) it is possible to browse product examples and design strategies that perform similar functions. This is claimed to enable users to explore different problem-solving strategies regarding a particular challenge or to get inspiration while browsing those strategies (Deldin & Schuknecht, Reference Deldin, Schuknecht, Goel, McAdams and Stone2014). Moreover, Cheong and Shu (Reference Cheong and Shu2013) suggest that providing multiple examples can enable the designer to recognize shared principles at a more abstract level because a variety of examples will not share the same superficial similarities.
3.3. Mode of representation: Displaying text, drawings, photographs, and so forth
Whether describing concrete examples or abstract principles, tool developers can present information in various ways by using text, drawings, photographs, animations, diagrams, equations, graphs, and other modes of representation (e.g., Chakrabarti et al., Reference Chakrabarti, Sarkar, Leelavathamma and Nataraju2005; Goel et al., Reference Goel, Vattam, Wiltgen and Helms2012). Some modes of representation are more effective than others for certain creative tasks, affecting the quantity, quality, and diversity of ideas generated (Fischer, Reference Fischer, Gero and Maher1993; Yamamoto & Nakakoji, Reference Yamamoto and Nakakoji2005; Sarkar & Chakrabarti, Reference Sarkar and Chakrabarti2008; Cardoso & Badke-Schaub, Reference Cardoso and Badke-Schaub2011). Combining different modalities, such as augmenting textual descriptions of an organism with diagrams of function structures, can improve the ability of designers to produce novel solutions (Linsey et al., Reference Linsey, Wood and Markman2008). Therefore, beyond just selecting the most appropriate mode of representation, providing the content in various forms enables the user to explore the content from multiple perspectives, thus maintaining a holistic perspective that promotes creativity (Candy & Edmonds, Reference Candy and Edmonds1995, Reference Candy and Edmonds1997; Hewett & DePaul, Reference Hewett, DePaul, Houstis, Rice, Gallopoulos and Bramley2000; Avital & Te'eni, Reference Avital and Te'eni2009).
3.4. Open-endedness: Permitting different interpretations of the information
Related to issues of abstraction, exemplification, and representation is the issue of how open-ended the content is. This involves encouraging users to perceive the content as being ambiguous, incomplete, uncertain, tentative, provisional, or partial, leading to further exploration and meaning-making activities (Candy & Edmonds, Reference Candy and Edmonds1996; Gaver et al., Reference Gaver, Beaver and Benford2003). If, instead, content is perceived as closed, definite, and finished, it is hard for users to conduct further exploration because the material is seen as an answer rather than a point of departure, which can result in fixation (Diggins & Tolmie, Reference Diggins and Tolmie2003; Töre Yargın, Reference Töre Yargın2013). The degree of open-endedness that is provided in the stimuli should be specified by considering the type of analogical design process: “while precise comprehension of biological systems can be ideal for detailed designs, for the purpose of concept generation, ambiguity and ‘generalness’ can also be useful” (Mak & Shu, Reference Mak and Shu2008, p. 28). Leaving content open to interpretation can be achieved through the judicious use of abstraction and exemplification and through selecting modes of representation that do not overdetermine the interpretations that are possible.
3.5. Concision: Balancing brevity with completeness
In providing information on the source domain, tool developers must decide how concise that information should be and how brevity should be balanced against completeness (Töre Yargın, Reference Töre Yargın2013). It is well known that many designers, especially in the early stages of design, have a tendency to overlook exhaustive reports in favor of more succinct guidance (Ramey et al., Reference Ramey, Robinson, Carlevato and Hansing1992; Kuniavsky, Reference Kuniavsky2003; Bartocci et al., Reference Bartocci, Potts and Cotugno2008). For the sake of system clarity, giving excessive information should be avoided because this may distract from the core underlying principles being communicated or may discourage sufficient browsing of multiple items (Diggins & Tolmie, Reference Diggins and Tolmie2003; Nørgaard & Hornbæk, Reference Nørgaard and Hornbæk2009). In contrast, the information should provide sufficient context so that the principle or example can be understood (Diggins & Tolmie, Reference Diggins and Tolmie2003; Blomberg & Burrell, Reference Blomberg, Burrell, Sears and Jacko2008). Providing multiple levels of detail is one strategy to overcome the competing objectives of providing information that is concise yet complete. However, multiplicity of information is a requirement that applies to more than the degree of concision and so is considered separately in the next section.
3.6. Multiplicity: Maintaining diversity and variety of content
From the five preceding subsections (Sections 3.1–3.5), it is clear that information from the source domain can be provided in many different ways. It is not necessarily the case that tool developers must make definitive decisions on each of these because they can choose to provide information in multiple ways concurrently. For instance, information can be provided at multiple levels of abstraction, from detailed examples to overall general principles, so that the user can identify analogical similarity at these different levels (Avital & Te'eni, Reference Avital and Te'eni2009; Goel et al., Reference Goel, Vattam, Wiltgen and Helms2012). Developers can provide multiple examples of entities that exhibit a similar property, so that users can understand the abstracted principles better or so that they can perform that abstraction themselves (Hewett & DePaul, Reference Hewett, DePaul, Houstis, Rice, Gallopoulos and Bramley2000; Deldin & Schuknecht, Reference Deldin, Schuknecht, Goel, McAdams and Stone2014). Different levels of abstraction can also be provided to allow users to identify patterns, commonalities, and anomalies that are not obviously seen at any one level (Avital & Te'eni, Reference Avital and Te'eni2009) and to thus more readily encourage the construction of useful analogies for design (Chakrabarti et al., Reference Chakrabarti, Sarkar, Leelavathamma and Nataraju2005; Goel et al., Reference Goel, Vattam, Wiltgen and Helms2012). Similarly, multiple modes of representation and multiple degrees of open-endedness and concision are possible.
4. INTERACTION QUALITIES PROVIDED BY THE TOOL
Beyond providing the right information in the right forms, it is important that tool developers consider the qualities of the interactions that users will have with that information. Developers must decide on how information can be accessed (e.g., searched or browsed), to what extent it can be modified or rearranged, and in what ways it can be shared. They must also consider how transparent the tool is and how well it connects with other systems or processes that the designer is using. These kinds of requirements for interaction are discussed here before their relationships with each other, with the information content requirements and with the intended outcomes are discussed in Section 5.
4.1. Accessibility: Allowing easy retrieval of the content
Tool users might have a clear idea what they are looking for and wish to search the database for keywords that they know to be relevant to their problem or solution. To prevent the list of search results from becoming unmanageable, developers can implement model-based tagging that assists with “information foraging” (Vattam & Goel, Reference Vattam and Goel2011), categorized search lists (Shu, Reference Shu2010; Cheong & Shu, Reference Cheong and Shu2012), or interactive overviews of categorized search results (Kules, Reference Kules, Shneiderman, Fischer and Hewett2005; Kules & Shneiderman, Reference Kules and Shneiderman2008). Such categorization of the data also permits browsing rather than searching, where users might accidentally find information that they could not have known to search for (Vattam & Goel, Reference Vattam and Goel2011). Different principles of classification are possible, but function-based classifications are most commonly adopted in bio-inspired design tools (see Section 3.1). These ontologies traditionally emphasize the function, behavior, and structure (or state) of the entities that are being classified or related. Other classification schemes are possible, however, and recent work has suggested that classification might also be conducted according to concepts of operating environment, performance criteria, deficiencies, benefits, constraints, or specifications (Helms & Goel, Reference Helms and Goel2013).
4.2. Interactivity: Providing active control and continuous feedback
Although the importance of interactivity is not commonly referred to in the analogical design support literature, its importance is evident in the literature on supporting creativity. Searching and browsing are ways in which users interact with the information content, but other forms of interaction are also possible. Interactivity is a complex quality, the definition of which is not commonly agreed on (Liu & Shrum, Reference Liu and Shrum2002; Johnson et al., Reference Johnson, Bruner, II and Kumar2006). However, there are some recurring constructs in operational definitions of interactivity, including “active control” and “reciprocity in communication.” These can be considered as common dimensions, which support exploration by enabling the tool to “talk back to the users” (Resnick et al., Reference Resnick, Myers, Nakakoji, Shneiderman, Pausch, Selker, Eisenberg, Shneiderman, Fischer and Hewett2005). To achieve this, it is important to provide continuous feedback and to maintain a sense of directness in interaction. Direct manipulation interfaces can facilitate interactive search queries by providing instant categorizations and visualizations to interpret the findings of research outcomes (Shneiderman, Reference Shneiderman1997), and such interactions can also support exploration of the content of analogical design tools.
4.3. Transparency: Providing clarity in interaction
While interacting with a tool, users may form an image of the way in which it works and the way in which they should use it. One way to assist in this is to transparently represent the tool's internal processes, so that users can feel that they are directly interacting with the process itself, promoting uptake of the tool (Edmonds et al., Reference Edmonds, Weakley, Candy, Fell, Knott and Pauletto2005). However, the tool should learn the user's language and practices, not the other way around (Fischer, Reference Fischer, Gero and Maher1993), allowing users to spend their resources on more valuable design activities (Resnick et al., Reference Resnick, Myers, Nakakoji, Shneiderman, Pausch, Selker, Eisenberg, Shneiderman, Fischer and Hewett2005; Avital & Te'eni, Reference Avital and Te'eni2009). Therefore, developers should identify suitable “black boxes” or “building blocks,” which enable the user to interact with the system to carry out design-related tasks, while disguising any underlying mechanisms that are irrelevant to those tasks (Resnick et al., Reference Resnick, Myers, Nakakoji, Shneiderman, Pausch, Selker, Eisenberg, Shneiderman, Fischer and Hewett2005). Web-based environments and typical web-based interaction styles are one approach to providing an interface that most computer-literate people can use, thus promoting exploration and uptake (Deldin & Schuknecht, Reference Deldin, Schuknecht, Goel, McAdams and Stone2014). Web-based environments also raise the possibility of easily connecting analogical design support tools with other systems and processes that designers engage in.
4.4. Connectivity: Integrating the tool with other systems
During the design process, designers might use different systems to locate information, document their thoughts, and record the progress made (e.g., computer-aided design software, digital sketching, and prototyping tools). Interoperability between these environments has proved important for supporting design activities (Shneiderman, Reference Shneiderman2000; Resnick et al., Reference Resnick, Myers, Nakakoji, Shneiderman, Pausch, Selker, Eisenberg, Shneiderman, Fischer and Hewett2005; Avital & Te'eni, Reference Avital and Te'eni2009). Similarly, having an integrated way of working with analogical design support tools would be beneficial, especially because the use of analogical design support tools might prompt further information searches outside the tool itself. For example, in response to locating an item of information in the tool, users may search the Internet, digital libraries, and other databases. Integrating with these other systems might involve allowing modifications to the tool itself and the addition of user-generated content found elsewhere (Hewett & DePaul, Reference Hewett, DePaul, Houstis, Rice, Gallopoulos and Bramley2000; Avital & Te'eni, Reference Avital and Te'eni2009). To achieve this, collaboration features that allow peer production of content are proposed for both the AskNature (Deldin & Schuknecht, Reference Deldin, Schuknecht, Goel, McAdams and Stone2014) and DANE tools (Goel et al., Reference Goel, Vattam, Wiltgen and Helms2012).
4.5. Shareability: Allowing content to be communicated to other stakeholders
Design processes often involve collaboration, even more so when the proposed designs draw on information or solutions from outside the designers' expertise (e.g., when drawing from the biological domain). Consequently, it is important that the content of the tool and the outcome of the user's progress should be shareable with other experts and stakeholders (Hewett & DePaul, Reference Hewett, DePaul, Houstis, Rice, Gallopoulos and Bramley2000; Edmonds et al., Reference Edmonds, Weakley, Candy, Fell, Knott and Pauletto2005; Avital & Te'eni, Reference Avital and Te'eni2009). This “shareability” of information supports effective communication within the design team and between the design team and others, thus promoting collaboration (Erickson, Reference Erickson, Luff, Hindmash and Heath1998; Hughes et al., Reference Hughes, O'Brien, Rodden, Rouncefield and Viller2000). However, the level of detail that designers require for communication may differ from the level of detail at which they are working (Edmonds et al., Reference Edmonds, Weakley, Candy, Fell, Knott and Pauletto2005). For example, a software programmer may want to work with symbolic codes but find it more productive to share information and ideas with others using visual representations. Therefore, analogical design support tools might provide representations of information in different modes, not only to enable analogy building, but also to enable analogy sharing.
4.6. Restoration: Permitting the resumption of previous activities
The importance of storing and returning to earlier states and activities is commonly emphasized for supporting work efficiency during creative activities (Elam & Mead, Reference Elam and Mead1990; Hewett & DePaul, Reference Hewett, DePaul, Houstis, Rice, Gallopoulos and Bramley2000; Shneiderman, Reference Shneiderman2000). The ability to store the history of a work session and then return to different points in that history can support designers in attending to other tasks and deferring judgments, both of which can support divergent thinking (Elam & Mead, Reference Elam and Mead1990). The restoration of prior states can be assisted by tool functionality that permits the naming, storing, and sorting of search sessions. An alternative approach is to provide features for browsing sessions to be recorded, retrieved, and reviewed. Whether by search or browse, it can be useful to allow designers to document their thoughts and decisions, so that they and other stakeholders can retrieve the rationale for the design decisions that they make (Bracewell et al., Reference Bracewell, Wallace, Moss and Knott2009). In order to not unnecessarily disrupt thought processes during these activities, the interface should permit flexibility in generating, storing, and retrieving ideas (Hewett & DePaul, Reference Hewett, DePaul, Houstis, Rice, Gallopoulos and Bramley2000).
4.7. Adaptability: Allowing the nature of the interactions to change
Analogical design tools should be adaptable and customizable for different users (e.g., expert vs. novice) and different working styles (e.g., systematic approaches vs. or inspiration-driven styles; Resnick et al., Reference Resnick, Myers, Nakakoji, Shneiderman, Pausch, Selker, Eisenberg, Shneiderman, Fischer and Hewett2005). In addition, there can also be different tasks to which the tool is applied, each of which might be better suited to different modes of interaction with the tool (Töre Yargın & Erbuğ, Reference Töre Yargın, Erbuğ, Bohemia, Liedtka and Rieple2012). This adaptability can be implemented manually, with the user able to detect which features correspond to a given working style (Hewett & DePaul, Reference Hewett, DePaul, Houstis, Rice, Gallopoulos and Bramley2000; Avital & Te'eni, Reference Avital and Te'eni2009). Alternatively, the tool can learn the user's pattern of working, adapting the tool to fit and automating certain tasks to improve work efficiency (Avital & Te'eni, Reference Avital and Te'eni2009). Moreover, the tool should be adaptable to different problem definition and reframing needs. For example, Chakrabarti et al. (Reference Chakrabarti, Sarkar, Leelavathamma and Nataraju2005) suggest two different ways of exploring the content of their IdeaINSPIRE tool based on how well the design problem can be defined by the user. Likewise, Helms et al. (Reference Helms, Vattam and Goel2009) define problem-based and solution-based processes for biologically inspired design and suggest features in the DANE tool to support these processes (Goel et al., Reference Goel, Vattam, Wiltgen and Helms2012).
5. THE RELATIONSHIPS BETWEEN REQUIREMENTS AND OUTCOMES
Developing tools that satisfy the user requirements for information content and interaction quality promises to deliver direct and indirect positive outcomes for the analogical design process. The requirements have complex relationships with each other and with the desired outcomes, as illustrated in Figure 2.
Fig. 2. Matrices illustrating the relationships between outcomes (rows) and requirements (columns). The numbers in the cells of the rectangular matrix indicate the relevant literature discussing the relation between a specific outcome and a specific requirement. The symbols in the cells of the triangular matrix indicate the relationships between requirements.
In the triangular matrix in Figure 2, we see that the relationships between requirements can be conflicting, indicating that trade-offs must be made; jointly required, indicating that they should be maintained together; and supporting, indicating that satisfaction of one positively impacts the other. Close inspection of Figure 2 reveals that two of the requirements (open-endedness and accessibility) are more heavily supported than the others. This can be interpreted as those other requirements serving as the means by which open-endedness and accessibility are achieved, which are in turn the means by which the positive outcomes are achieved. Having identified open-endedness and accessibility as key requirements, we can sort the rest of the requirements into three groups based on their relationships with open-endedness and accessibility: requirements supporting both open-endedness and accessibility, conflict generating requirements (those that generate conflicts between open-endedness and accessibility), and conflict resolving requirements (those that resolve the conflicts generated between open-endedness and accessibility). In Figure 3, these groups of requirements are illustrated in a network diagram to better represent the relationships between the requirements and how these might be considered during tool development. Each of these relationships is discussed in turn below.
Fig. 3. Network diagram of the supporting and conflicting relationships between individual requirements.
5.1. Requirements supporting both open-endedness and accessibility
Representation, interactivity, and abstraction are the three requirements that each support both of the key requirements. Certain modes of representation are more abstract, ambiguous, or multilayered than others; and such modes are thus more open to interpretation, permitting designers to explore (and reexplore) the content from different perspectives. Selecting the right mode of representation can also make the content more accessible by quickly revealing the most salient features while still permitting deeper exploration (Biomimicry 3.8 Institute, 2008–2014). Enhancing a tool's interactivity means increasing the degree to which it provides continuous, reciprocal, and immediate feedback. By allowing users to work with, rather than just read from, the required content, different ways of viewing the content can be explored, making it more accessible. Current tools can benefit from direct manipulation mediums and interactive search facilities that can provide instant categorizations and visualizations (Shneiderman, Reference Shneiderman1997). Through abstraction of the content, classification is implied, which makes the content more accessible through search and browse activities (Chakrabarti et al., Reference Chakrabarti, Sarkar, Leelavathamma and Nataraju2005; Vincent et al., Reference Vincent, Bogatyreva, Bogatyrev, Bowyer and Pahl2006; Shu, Reference Shu2010; Vattam & Goel, Reference Vattam and Goel2011; Cheong & Shu, Reference Cheong and Shu2012; Goel et al., Reference Goel, Vattam, Wiltgen and Helms2012; Deldin & Schuknecht, Reference Deldin, Schuknecht, Goel, McAdams and Stone2014). Abstracted principles are also more open-ended because they eliminate details to represent more general principles (Mak & Shu, Reference Mak and Shu2008).
5.2. Conflict generating requirements
Exemplification, multiplicity, connectivity, concision, transparency, and shareability all conflict with one of the two key requirements and yet are also important to delivering the required positive outcomes. This requires tool developers to make trade-offs in balancing the requirements they satisfy. Exemplification is important to clarify abstracted principles, but in so doing, makes those principles less general and risks inducing fixation, thus conflicting with open-endedness. To avoid this, surface-dissimilar, unusual, and multiple varied examples should be provided (Nijstad et al., Reference Nijstad, Stroebe and Lodewijkx2002; Perttula & Sipilä, Reference Perttula and Sipilä2007; Wilson et al., Reference Wilson, Rosen, Nelson and Yen2010). In that sense, multiplicity supports open-endedness of the content and can help designers to avoid or overcome fixation. Connectivity also supports open-ended interpretation of the content by allowing the designer to access different sources or systems and to connect the information in new ways. Search tools based on a natural-language approach that can be connected to different sources can potentially allow designers to access a diversity of sources, thus facilitating more open-ended discovery (Shu, Reference Shu2010). Despite their benefits, multiplicity and connectivity both conflict with concision and accessibility, because they broaden and complicate the possible sources and formats of material available. Categorizing the search outcomes and filtering the irrelevant and frequently appearing words are some of the strategies that are adopted to deal with this conflict (Cheong & Shu, Reference Cheong and Shu2012). Transparency enables the user to understand how to interact with the tool, and thus supports accessibility, but open-endedness can conflict with this requirement, because making the content open to interpretation can reduce clarity. In addition, making the content shareable between the stakeholders in the team increases its accessibility for team members, whereas making it open-ended and interpretable can reduce its shareability. This is because everybody who uses the tool has his or her own interpretations, and it can thus be hard to have shared understanding within the team's communication (Stacey & Eckert, Reference Stacey and Eckert2003).
5.3. Conflict resolving requirements
Adaptability and restoration are requirements that can resolve the conflicts discussed above. Maintaining diversity in the content and connecting the tool with other systems can jeopardize its accessibility. One way to resolve this is to make the tool adaptable by providing ways to filter the information or change the mode of interaction based on user needs and preferences (e.g., whether for problem-oriented or solution-oriented design processes in Goel et al., Reference Goel, Vattam, Wiltgen and Helms2012). Likewise, through adaptability, the system can be made more transparent because the irrelevant content can be flexibly filtered out, preventing it from obscuring the information and interactions that are necessary. In addition, adaptability can make the content more shareable because that content can then be modified for the needs and preferences of different of users. Another way to overcome the challenges to accessibility is to implement restoration, thus making prior states more accessible, potentially for different users (Hewett & DePaul, Reference Hewett, DePaul, Houstis, Rice, Gallopoulos and Bramley2000).
As in other areas of design, identifying and reconciling conflicting requirements in tool development is important for innovation, and developers should attend to the conflicts highlighted above. However, other relationships between the requirements might also be relevant, and this should be investigated in future work.
6. DISCUSSION
In reviewing user requirements for tools that support analogical design, a number of avenues for future work emerge. This work would provide further guidance on how such tools should be developed by better understanding those requirements that are conflicting, poorly defined, or poorly supported. For example, considering the requirements for information content, the issue of abstraction raises a number of research issues. Abstraction is one of the key principles for analogical transfer, but it is important to identify the right principles upon which to perform the abstraction. To date, function has been the most common principle of abstraction and classification. However, function is a problematic concept (e.g., Crilly, Reference Crilly2010; Vermaas & Eckert, Reference Vermaas and Eckert2013), and other complementary principles of abstraction might be used, including those relating to system architectures such as modularity, redundancy, robustness, and scalability (e.g., Hastings & McManus, Reference Hastings and McManus2004; Knippers & Speck, Reference Knippers and Speck2012; Chen & Crilly, Reference Chen and Crilly2014). Future work could usefully identify the variety of abstraction principles that might be applied and seek to understand how they might most effectively be selected from or combined.
Considering the requirements for interaction qualities, the related issue of accessibility is also important. This is because the searching and browsing activities by which content is accessed influences, or is even determined by, the principles upon which that content is abstracted. Furthermore, in order to develop tools that integrate with the other systems that designers use, providing connectivity is important. Better understanding the requirement for connectivity requires research into the variety of existing tools, systems, and processes that are relevant to analogical design and research into the real-world contexts in which those tools are used and in which analogical design takes place. In order to investigate their expectations, past experiences of designers who have expertise in analogical design should be consulted to understand their information and interaction needs, while accessing source analogs and mapping and transferring from source analogs to the target solution. Identifying these criteria is necessary both for developing future tools and for evaluating them. This would yield guidance on the standards, conventions, and terminology that should be adopted within and across tools.
Analogical design support tools can serve designers in some of the most difficult challenges in design: the identification of principles and precedents that are related to (however distantly) the design task under consideration. For developers striving to provide such tools, there are a number of activities that must be undertaken. They must generate a broad and detailed catalog of possible sources for designers to draw from; they must determine how that information should be structured and presented; and they must also implement an interface to that information that promotes interactions that are engaging and effective. These activities should satisfy the requirements of the tools' users, that is, the designers who will work with them. Developing design support tools according to those requirements promises to decrease development time and also to increase uptake. We have sought to gather such requirements in this paper and have explored the relationships between them. Conducting further work into these requirements and their relations would provide a more solid foundation upon which analogical design support tools can be developed and deployed, and assist in realizing their potential in design research, practice, and education.
ACKNOWLEDGMENTS
Gülşen Töre Yargın's work was supported by the International Post Doctoral Research Fellowship Programme (BİDEB-2219) from the Scientific and Technological Research Council of Turkey. Nathan Crilly's work was supported by an Early Career Fellowship (EP/K008196/1) from the United Kingdom's Engineering and Physical Sciences Research Council.
Gülşen Töre Yargın is a Visiting Postdoctoral Research Associate in the Engineering Design Centre at the University of Cambridge. She received her PhD from the Middle East Technical University, Department of Industrial Design. She holds BID and MS degrees. Her research interests focus on communication of research knowledge in the design process. She is particularly interested in the design and development of research deliverables and information visualization. Dr. Töre Yargın has more than 6 years of research experience at METU/BILTIR-UTEST Product Usability Unit as a User Researcher on concepts ranging from consumer products to automotive design.
Nathan Crilly is a Senior Lecturer in engineering design at the University of Cambridge. He has a PhD in design research and a Bachelors degree in mechanical engineering. His research interests are in the areas of design, technology, and communication. He employs an interdisciplinary approach to studying how artifacts (e.g., products, systems, or services) are developed, the properties they exhibit, and the ways in which people use them. Nathan is a Fellow in Engineering at Clare College, Cambridge. Dr. Crilly is a member of the Design Research Society and the Design Society. He also serves on the International Editorial Board of Design Studies.
