Cirp Journal of Manufacturing Science and Technology: A.M.M. Sharif Ullah, Md. Mamunur Rashid, Jun'ichi Tamaki

G Model
CIRPJ-183; No. of Pages 12
CIRP Journal of Manufacturing Science and Technology xxx (2011) xxx–xxx
Contents lists available at SciVerse ScienceDirect
CIRP Journal of Manufacturing Science and Technology

journal homepage: www.elsevier.com/locate/cirpj
On some unique features of C–K theory of design

A.M.M. Sharif Ullah *, Md. Mamunur Rashid, Jun’ichi Tamaki
Department of Mechanical Engineering, Kitami Institute of Technology, 165 Koen-cho, Kitami, Hokkaido 090-8507, Japan
A R T I C L E I N F O A B S T R A C T
Article history: Concept–Knowledge theory (C–K theory) of design is a relatively new theory for describing reasoning
Available online xxx and creative processes in engineering design. This paper describes some unique features of this theory. In
particular, it is shown that C–K theory encompasses logical inferences that are more complex than
Keywords: classical abduction. A design process in C–K theory is rather motivation-driven and this motivation can
C–K Theory be quantified by the concept called information content (entropy) measured under epistemic
Engineering design uncertainty. Since the Internet-driven information will play a major role in performing engineering
Abduction
design (building concept, acquiring domain knowledge, and alike) in the near future, the scope and
Motivation
Information content
limitation of building a C–K map by using the Internet is described. This provides some unexplored issues
Internet of engineering design.
Concept map ß 2011 CIRP.
1. Introduction co-creation of new knowledge) has been a subject of research. Some

of the relevant works are described below.
To describe reasoning and processes of innovation and creativity Kazakci and Tsoukias [11] showed how to develop design tools
in engineering design, mapping of objects from one domain (or for practicing creativity by using C–K theory. They have introduced
space) to another have been found effective. For example, consider a domain called Environment in addition to the domains of
the mappings (i) among Functions (F), Behaviors (B), and Structures Concept and Knowledge to achieve this. Hatchuel and Weil [9]
(S) introduced by Gero [7], (ii) between Functional Requirements have shown that stability of object in Knowledge domain explains
(FR) and Design Parameters (DP) introduced by Suh [21], and (iii) the topological structure of design modeling introduced by Braha
between Concept (C) and Knowledge (K) introduced by Hatchuel and and Reich [2]. Therefore, topological structure of design modeling
Weil [8,9]. This article deals with some unique features of C–K and C–K theory are synergistic to each other in dealing with the
mapping underlying the design theory called Concept–Knowledge structured knowledge for designing creative artifacts. Galle (2009)
theory (hereinafter referred to as C–K theory) [8,9]. A schematic [32] has found that Function–Behavior–Structure models [7] avoid
illustration of C–K theory is shown in Fig. 1. As seen from Fig. 1, there ‘‘as-yet non-existent’’ objects manifesting a phenomenon called
are two interdependent domains called Concept Domain and un-embodied structure. To fill this gap, the idea of undecided
Knowledge Domain in C–K theory of design. In addition, there are concept of C–K theory can be used (Galle 2009, [32] pp. 335–336).
mappings between C and K, i.e., C ! K, K ! C, C ! C, and K ! K. This Reich et al. [17] have critically analyzed the Advanced Systematic
mapping is somewhat different compared to those in other design Inventive Thinking (ASIT) (a derivative of the Altshuller’s Theory of
theories. For example, in Axiomatic Design [21] the mapping is Innovative Problem Solving [1]) and have shown that C–K theory
allowed in a hierarchical manner: FR ! DP ! FR (new) ! DP (new). subsumes innovation mechanism of ASIT. Hatchuel et al. (2011)
The mapping FR to FR or DP to DP is not allowed in Axiomatic Design. [10] have shown that C–K theory helps overcome ‘‘fixation
However, one of the most remarkable features of C–K mapping is its effects’’—the effects that hinder creativity. They have found that
ability to dealing with a creative concept—a concept that is the outcomes of C–K theory based design curriculum are
undecided with respect to the existing knowledge at the point of measurable—a desirable characteristic for educating students
time when it (the concept) is conceived. If such an undecided with the ability of creative thinking.
concept is pursued further, new knowledge might evolve in favor of There are other unique features of C–K theory that need
the concept. As a result, both knowledge evolved and concept investigations. For example, a question may be asked: Is it possible
conceived become the part of design. Such a remarkable feature to position C–K theory in terms of abduction (an important logical
of C–K theory (i.e., an ability to deal with undecided concepts and inference for formally incorporating creativity (at least innova-
tion))? In addition, since a concept is an undecided entity at the
beginning, there must a ‘‘motivation’’ behind pursing it further.
* Corresponding author. Tel.: +81 157 26 9207; fax: +81 157 26 9207. This raises a question: What is the nature of motivation involved in
E-mail address: ullah@mail.kitami-it.ac.jp (A.M.M. Sharif Ullah). pursuing an undecided concept? Moreover, questions may be
1755-5817/$ – see front matter ß 2011 CIRP.

doi:10.1016/j.cirpj.2011.09.001
Please cite this article in press as: Sharif Ullah, A.M.M., et al., On some unique features of C–K theory of design. CIRP Journal of
Manufacturing Science and Technology (2011), doi:10.1016/j.cirpj.2011.09.001
G Model
2 A.M.M. Sharif Ullah et al. / CIRP Journal of Manufacturing Science and Technology xxx (2011) xxx–xxx
Fig. 1. An illustration of C–K theory of design.
asked: Is it possible to measure the information content (or There are many forms of abduction [26,18]. One of the basic forms
entropy) of a concept even though it is undecided (i.e., under is as follows:
epistemic uncertainty)? How does the information content vary
while continuing a design process in accordance with C–K theory?
p→q
Is it possible to build a C–K map by using Internet-driven q (1)
information resources?
p (a possible outcome)
The remainder of this paper provides answers to the above
mentioned questions. The sections are organized, as follows:
Section 2 describes C–K theory in terms of abduction and The expression in (1) means that if the consequent (q) of a
motivation. Section 3 measures the information content of creative logical implication ‘‘p ! q’’ is true, then the antecedent (p) is a
and ordinary concepts from the view point of epistemic ‘‘possible’’ outcome. In other words, there might be other possible
uncertainty. Section 4 provides a discussion on the findings in outcomes in addition to p. Thus, abduction refers to multiple
Sections 2 and 3 and highlights the implication of building C–K outcomes. This is schematically illustrated in Fig. 2 using a logical
map using Internet-driven information under current information implication ‘‘bird ! fly.’’
retrieval technology. Section 5 concludes this study. A seen from Fig. 2, from the logical implication ‘‘bird ! fly,’’
‘‘bird’’ and ‘‘some other objects’’ are the outcomes when ‘‘fly’’ is the
2. Abduction, motivation and C–K theory requirement. If someone consults the knowledge of objects able to
fly, he/she could find that an object called ‘‘helicopter’’ (for
2.1. Abduction example) is consistent with the knowledge. Thus, instead of the
solution called ‘‘bird,’’ another solution called ‘‘helicopter’’ might
Providing a logical explanation of a cognitive process of design be adopted as a design solution (an object that can fly), if it appears
has been an active area of research. Many authors have studied this to be more appropriate for a given situation. A more human-
issue using different approaches. For example, see the works of friendly representation of abduction-based design process (Fig. 2)
Yoshikawa [29], Zeng and Cheng [30], Kazakci et al. [12], is illustrated in Fig. 3. The illustration in Fig. 3 is actually a concept
Tomiyama et al. [23], and Ullah [26]. Some of the authors have map.1 Thus, this concept map in Fig. 3 is a visual representation of
identified that the logical inferences, namely, deduction, induction,
1
and abduction, are associated with the cognitive processes of Concept maps are graphical representation of entities and their relationships.
Using concept map one can create a ‘‘meaning base’’ of an issue. The authors use
design. Particularly, abduction (opposite to deduction) is consid- concept maps throughout this paper to illustrate C–K map and other related
ered an important ingredient for dealing with the creativity (at processes. To know the details of concept map and its computing tools, refer to refs.
least innovation) while continuing a design process [29,23,26]. [15,3,14].
G Model
A.M.M. Sharif Ullah et al. / CIRP Journal of Manufacturing Science and Technology xxx (2011) xxx–xxx 3
Fig. 2. Implication of abduction.
the typical nature of continuing a design process by using classical

abduction.
Even though abduction helps come up with solutions other
than the obvious ones, the solutions should be consistent with the
existing knowledge or modified knowledge without considering
new facts (see [26] for more details). This is perhaps a special case
in C–K theory, as mentioned by Hatchuel and Weil [9] while
positioning it (C–K theory) with respect to the work of Braha and
Reich [2]—an advanced version of general theory of design [29]
driven design process. The general case is that the proposed
solution may not have to be consistent with the existing
knowledge. Thus, C–K theory puts a design process beyond the
scope of classical abduction-driven logical inference. Such Fig. 3. A concept map of abduction-driven design process.
generality is explained by another concept map shown in
Fig. 4. The concept map in Fig. 4 cannot be expressed the
diagram like it is in Fig. 2 or even by a logical expression like it is performance of C2 is not clearly known because the knowledge of
in (1). the performance of C2 (K4) is empty, i.e., K4 = {1}. Thus, a
As seen from Fig. 4, in C–K-theory-based design process, one challenge of seeking new knowledge emerges. Overcoming this
can propose a solution X (i.e., a concept) that is undecided (i.e., challenge acts as the other motivation for pursing C2 instead of C1.
cannot be rejected or accepted) with respect to the existing This motivation is hereinafter referred to as epistemic challenge.
knowledge. Thus, if there is enough ‘‘motivation,’’ then efforts can Thus, the interplay of compelling reason and epistemic
be made to explore new knowledge and after exploring new challenge creates a scope that actually helps pursue C2, not C1.
knowledge if it (new knowledge) supports X, then X becomes a In general, a concept like C1 should entail low degree of compelling
solution. Therefore, cognitive processes involved in C–K-theory- reason and epistemic challenge (i.e., low motivation), whereas a
based design process are driven by a special kind of ‘‘motivation’’ concept like C2 should entail high degree of compelling reason and
not by classical abduction as such. Now the question is: what is the epistemic challenge (i.e., high motivation). Now the question is:
nature of such motivation and how to quantify it? The qualitative how to measure the degrees of compelling reason and epistemic
nature of this motivation is described in the following subsection. challenge? An answer to this question is to measure the entropy or
A quantitative nature is described in the following section. information content of the concepts from the view point of
epistemic uncertainty (i.e., uncertainty due to lack of knowledge,
2.2. Nature of motivation preference, difference in opinion, and alike).
The concept of information content formally appears in the
As mentioned before, motivation drives the design process literature of design theory as a result of the introduction of Second
toward conceiving X (an undecided concept as illustrated in Fig. 4). Axiom of Axiomatic Design Theory of Systems [21]. To measure such
In fact, recent research works show that motivation is one of the information content, it is necessary to have two pieces of
key factors behind designing creative artifacts [22]. In particular, information: (1) system range and (2) design range. The information
Chakrabarti [4] has found that motivation is an entity that is able to content then measures the degree of match between system range
trigger knowledge (or vice versa). This interplay activates other and design range. Note that design range is the requirement that is
entities, namely, ‘‘effort’’ and ‘‘ability.’’ As a result, creativity comes set by the designer and system range is what a system (i.e., proposed
into being (see [4, p. 50]). design) offers. As such, information content of a design from the
However, to identify the nature of motivation behind pursuing sense of ‘‘degree of match between system range and design range’’
an undecided concept like X, consider two concepts: an ordinary is the information content of requirement (hereinafter referred to as
concept C1 (=fossil-fuel-based propulsion engine) and a creative requirement information content or requirement entropy). It is
concept C22 (=Mg–CO2-based propulsion engine). The underlying worth mentioning that, in practice, design range is expressed by a
C–K map is shown in Fig. 5. Given the knowledge of Earth crisp range and system range is expressed by a probability
atmosphere (K1) and Mars atmosphere (K2), it is perhaps true that distribution. This way, the requirement information content, as it
C1 is a suitable solution for Earth not for Mars, whereas C2 is has been practiced, actually helps measure the aleatory uncertainty
perhaps a suitable solution for Mars not for Earth. This is one of the of design. In addition, an analysis of requirement information
motivations behind pursuing C2 instead of C1 for Mars exploration. content can be carried out only when a designer has a great deal of
This motivation is hereinafter referred to as compelling reason. One analytical knowledge (e.g., FR–DP mapping is fully defined by some
the other hand, the performance of C1 can be determined by using algebraic (analytical) expressions).
the existing knowledge of engine performance (K3), whereas the On the other hand, under epistemic uncertainty it is not
possible to define the structures and processes in terms of pure
2
C2 is the same concept used by Hatchuel and Weil [9] while proposing an analytical expressions, and, thereby, to practice the above
advanced formulation of C–K theory. mentioned requirement information content. To deal with
G Model
Fig. 4. Concept map of C–K theory based design process.
Fig. 5. A concept map of creative concept based on C–K theory.
G Model
Table 1
Setting for determining the information content of C1.
Propositions Truth values Requirement
P11 C1 is suitable for Mars atmosphere Mostly false 0.1 Engine should be suitable for Mars atmosphere
P12 C1 is not suitable for Mars atmosphere Perhaps true 0.73
P13 Performance of C1 is satisfactory Mostly true 0.9 Engine performance should be satisfactory
P14 Performance of C1 is not satisfactory Mostly false 0.1
epistemic uncertainty the concepts called possibility (e.g., fuzzy considered. This way, Certainty Entropy equal to 0 means that
logic based approaches), plausibility (e.g., evidence theory based all propositions are either completely true or false. This happens
approaches), and alike can be used. In possibility based when the knowledge is complete, i.e., the issue is clearly known.
approaches, in particular, the degree of possibility (truth value Certainty Entropy is equal to 1 means that all propositions are
of proposition or degree of membership of an entity with respect to equally true and false (i.e., all truth values are equal to 0.5). This
a linguistically defined entity in the scale of [0, 1]) plays the key happens when the knowledge is absolutely incomplete (a state of
role [31]. It is worth mentioning that fuzzy number (i.e., possibility complete ignorance). On the other hand, Requirement Entropy is
distribution in terms of degree of membership or truth value) is a rationalized by the Desire Definiteness Axiom. In case of epistemic
probability-distribution-free representation of uncertainty [5]. uncertainty, the requirement (desire or design range) is defined by
This means that a fuzzy number encodes a family of probability a linguistic proposition. As such, Requirement Entropy is equiva-
distributions. Therefore, when there is no information available for lent to requirement information content. Requirement Entropy is
defining system range or design range in terms of a crisp range or also measured in the scale of 0–1, wherein 0 means complete
probability distribution (i.e., under epistemic uncertainty), it is a fulfillment of requirement and 1 means the opposite. Therefore,
convenient option to perform formal computation by using fuzzy under epistemic uncertainty, the information content of a concept
numbers or truth value based approaches. with respect to an issue has two dimensions: Certainty Entropy
and Requirement Entropy. The relative positions of Certainty
2.3. Measuring motivation Entropy and Requirement Entropy for both compelling reason and
epistemic challenge collectively measure the information content
As explained in the previous subsection, measuring the degree of concepts like C1 and C2.
of motivation in terms of compelling reason and epistemic
challenge under epistemic uncertainty can be achieved by using 3. Information content and C–K theory
possibility based approach, i.e., by assigning truth values to a set of
propositions or degrees of membership to a set of linguistic classes. 3.1. Information content of C1
Since the state is epistemic uncertainty, there should be clear
procedure to capture and quantify the lack or abundance of To measure the information content of C1 with respect to C–K
knowledge. At the same time, there should be a clear procedure to map in Fig. 5 in terms of Certainty Entropy and Requirement
compute and quantify the degree of match between system range Entropy, two sets of propositions {P11, P12} and {P13, P14} are
and design range, although these ranges are not clearly defined. To considered. The first set, {P11, P12}, deals with the motivation
achieve this, four axioms have been introduced, as follows: Local called compelling reason, i.e., whether or not it is true that C1 is
Definiteness Axiom, Global Definiteness Axiom, Granule Definite- suitable for Mars exploration. Thus, the underlying design
ness Axiom, and Desire Definiteness Axiom [24,25]. The details requirement should be ‘‘suitable engine’’ for Mars atmosphere.
descriptions of these axioms can be found in [24,25]. These axioms On the other hand, the other set, {P13, P14}, deals with the
provide numerical measures that help quantify information motivation called epistemic challenge, i.e., whether or not it is true
content or entropy under epistemic uncertainty. See Ullah that C1’s performance is known. The underlying design require-
[24,25], Ullah et al. [28], Ullah and Harib [27] for more details ment this time is ‘‘satisfactory performance’’ of an engine. Table 1
of the measures. Appendix A provides a summary of these summarizes {P11, P12}, {P13, P14}, and the design requirements.
measures customized for this study. Referring to the mathematical As listed in Table 1, P11 is ‘‘mostly false,’’ P12 is ‘‘perhaps true’’
expressions as shown in Appendix A, the semantics of these given the knowledge of Mars atmosphere (K2). On the other hand,
measures are described, as follows: P13 is ‘‘mostly true’’ and P14 is ‘‘mostly false,’’ given the knowledge
Information content3 (I(TV)) of truth value of a proposition (TV) of engine performance (K3). The values of Certainty Entropy and
is rationalized by the Local Definiteness Axiom. This simply means Requirement Entropy of {P11, P12} are 0.37 and 1, respectively,
that if the truth value is 0 (proposition is completely false) the whereas, the values of Certainty Entropy and Requirement Entropy
information content is 0 (complete order). If the truth value is 1 of {P13, P14} are 0.2 and 0, respectively. The overall information
(proposition is completely true) the information content is 0 content of C1 is 1.74 (i.e., the value of coherency measure [28,27]).
(complete order). In other orders, truth value 0 or 1 implies that the The information contents are plotted in Fig. 6. Note the opposite
knowledge is complete—otherwise, it is not possible to say that the positions of epistemic challenge and compelling reason in Fig. 6.
proposition is completely true (truth value 1) or false (truth value Epistemic challenge has low information content (i.e., it is not a
0). If the truth value is 0.5 (proposition is equally true and false), challenge as such), whereas compelling reason has high informa-
the information content is 1 (complete disorder or huge lack of tion content (i.e., it is not serving as a compelling reason as such).
knowledge). For other cases, the information content is between 0
and 1. Certainty Entropy is rationalized by the Global and Granular 3.2. Information content of C2
Definiteness Axioms. It is measured by calculating the average
information content of the truth values of all propositions Similar to the previous case, to measure the information
content of C2 in terms of the C–K map in Fig. 5, two sets of
3
This information content (I(TV)) is just the information content of truth value
propositions {P21, P22} and {P23, P24} are considered. Table 2
(TV) of a proposition. It should not be considered the information content of design summarizes {P21, P22}, {P23, P24}, and the design requirements.
(i.e., requirement information content of design). As listed in Table 2, P21 is ‘‘perhaps true,’’ whereas P22 is ‘‘perhaps
G Model
Fig. 6. Information content of C1. Fig. 7. Information content of C2.
false,’’ based on Mars atmosphere (K2). On the other hand, both

P23 and P24 are ‘‘not sure’’ (i.e., neither true nor false) because K4
is empty, i.e., a complete lack of knowledge. Table 2 also
summarizes the settings for determining the information content
of C2 in terms of {P21, P22} and {P23, P24}.
The values of Certainty Entropy and Requirement Entropy for
{P21, P22} are 0.54 and 0, respectively, whereas the values
Certainty Entropy and Requirement Entropy for {P23, P24} are
both 1. The overall information content of C2 is equal to 3 (in terms
of coherency measure), which is a very high value compared to that
of C1 (1.74). Fig. 7 plots the information contents for C2. Note the
opposite positions of epistemic challenge and compelling reason in
Fig. 7. Epistemic challenge has very high information content (i.e.,
it is indeed a challenge), whereas compelling reason has low
information content (i.e., it is indeed a compelling reason).
3.3. Effectiveness of new knowledge
Lack of knowledge (K4 = {1}) actually results a very high

information content for C2, compared to that of C1, as it is seen
from the results in Figs. 6 and 7. If new knowledge explaining the
performance of C2 could have been gathered (i.e., K4 transforms to Fig. 8. Expected reduction in information content due to new knowledge.
K0 4), then the information content of C2 might have gone down
significantly. This expectation is schematically illustrated in Fig. 8.
To validate this expectation, a C–K map is constructed as shown in been used as an example. The sixth hit was a PDF document
Fig. 9 transforming K4 into K0 4. This time, the Concept Domain and containing a report from NASA written by Foote and Litchford [6].
Knowledge Domain are not shown separately. To formulate K0 4, a This report first summarizes the initial findings of some unusual
phrase called ‘‘Mars Exploration Mg–CO2 Engine’’ is used to fuel and oxidizer combinations reported by Shafirovich et al.
searching the Internet. The search engine called GoogleTM is used [19,20]. The C–K map in Fig. 9 provides an explicit representation
to perform the search. The last search was conducted on 9 March of such finding reported by Foote and Litchford [6].
2011 at 9:00 PM Japan Standard Time. There were 74 hits As seen from Fig. 9, for Mars exploration, a propulsion engine is
(websites) for the keyword ‘‘Mars Exploration Mg–CO2 Engine.’’ needed that should use in situ fuel and oxidizer. Given the fact that
First five hits deals with the C–K theory itself because in the Mars atmosphere consists of more than 95% CO2, it (CO2) can be
literature of the C–K theory Mg–CO2-based propulsion engine has used as an oxidizer, even if it is an unusual choice. This necessitates
Table 2
Setting for determining information content of C2.
P21 C2 is suitable for Mars atmosphere Perhaps true 0.73 Engine should be suitable for Mars atmosphere
P22 C2 is not suitable for Mars atmosphere Perhaps false 0.27
P23 Performance of C2 is satisfactory Not sure 0.5 Engine performance should be satisfactory
P24 Performance of C2 is not satisfactory Not sure 0.5
G Model
Fig. 9. C–K map when K4 transforms into K0 4.
a particular type of fuel either metals (Be, Mg, Al, Li, Ca, etc.) or their produces almost the same amount of Specific Impulse (an
hydrates (e.g., BeH2, MgH2, etc.). The fundamental studies important performance measure of propulsion devices) compared
conducted by Shafirovich et al. [19,20] reveals that the fuels, to that of other combinations (i.e., Al–CO2, Be–CO2, and BeH2–CO2).
namely, Mg, Al, Be, BeH2 are probably the most useful fuels when In terms of other important performance measures (i.e., combus-
CO2 acts as the oxidizer. It is also found that Mg–CO2 combination tion characteristics, such as toxicity, ignitability, combustion rate,
G Model
Table 3
Setting for determining information content of C2 using the C–K map in Fig. 9.
P1 C2 is acceptable in terms of Specific Impulse Perhaps true 0.73 An engine should be

P2 C2 is not acceptable in terms of Specific Impulse Mostly false 0.1 acceptable in terms
P3 C3 is acceptable in terms of Specific Impulse Mostly true 0.9 of Specific Impulse
P4 C3 is not acceptable in terms of Specific Impulse Mostly false 0.1
P5 C2 is acceptable in terms of toxicity, ignitability, combustion rate, slag formation, etc. Mostly true 0.9 An engine should be
P6 C2 is not acceptable in terms of toxicity, ignitability, combustion rate, slag formation, etc. Mostly false 0.1 acceptable in terms of
P7 C3 is acceptable in terms of toxicity, ignitability, combustion rate, slag formation, etc. Mostly false 0.1 toxicity, ignitability,
P8 C3 is not acceptable in terms of toxicity, ignitability, combustion rate, slag formation, etc. Perhaps true 0.73 combustion rate, slag
formation, etc.
C2 = Mg–CO2 propulsion engine; C3 = Y–CO2 propulsion engine, Y 2 {Be, BeH2, Al}.
and slag formation) Mg–CO2 combination produces relatively the point of time when it (the concept) is conceived) appears to be
better result. more complex than the logical inference called abduction [29,23,26],
To measure the information content of the concepts in Fig. 9, a see Fig. 4. It is rather a motivation driven process having two facets
set of propositions P1, . . ., P8 and two types of concepts, one is C2 called compelling reason and epistemic challenge. These facets of
(same as before) and the others are C3 (=Y–CO2-based propulsion motivation can be quantified under epistemic uncertainty using two
engine, Y is either Be or BeH2 or Al) are considered (see Table 3). It is types of entropies (or information contents) called Certainty Entropy
needless to say that the truth values of the propositions listed in and Requirement Entropy. While conceiving a creative concept, the
Table 3 reflect the facts shown in Fig. 9. information content should be maximized (compare the informa-
The information content in terms of Certainty and Requirement tion content of C1 and C2 as shown in Sections 3.1 and 3.2,
Entropies are determined by using the same methods used in the respectively). On the other hand, while pursuing a conceived
previous subsections. The results are plotted in Fig. 10. C2 has (creative) concept in presence of new knowledge, the information
information contents (0.37, 0) and (0.2, 0) for {P1, P2} and {P5, P6}, content should be minimized (compare the information content of
respectively (note the points in Fig. 10(a)). The overall information C2 and C3 in Section 3.3). Thus, the cognitive process of design based
content of C2 is equal to 0.57. On the other hand, C3 has on C–K theory can schematically be expressed by the illustration in
information content (0.2,0) and (0.37,1) for {P3, P4} and {P7, P8}, Fig. 11. As seen from Fig. 11, due to lack of knowledge, the
respectively (note the plot in Fig. 10(b)). The overall information information content of a concept jumps to its peak. At the same time,
content of C3 is equal to 1.74. Thus, if the designer chooses C2, not if motivating factors called compelling reason and epistemic
C3, the design process underlies ‘‘minimization of information challenge work, then the design process conceives a concept. When
content,’’ i.e., the design process complies with the Second Axiom a substantial amount of knowledge becomes available (gain of
of Axiomatic Design [21]. This is just the opposite compared to the knowledge), the information content of the conceived concept
cases shown in the previous two subsections. should go down significantly. In this case, the conceived concept
becomes a part of the design, i.e., the concept is adopted as a solution
4. Discussion to the design problem. Otherwise, the design process should
abundant the conceived concept and a new course of direction
The descriptions in the previous two sections reveal some unique should be explored. This way, the existence of iterations in design is
features of C–K theory. The design process underlying C–K theory explained by C–K theory.
(particularly, the process when a creative concept is conceived— Nevertheless, there are many computation challenges of
concept that is undecided with respect to the existing knowledge at building a C–K map. One of the computational challenges is
Fig. 10. Information content of C2 and C3 based the C–K map in Fig. 9.
G Model
Fig. 11. Unique characteristic of C–K theory based design process.
how to assist designers building a C–K map. Once a C–K map is

built, how to retain and reuse the outcomes is another challenge.
Some authors have suggested adding new domain called
Environment [11] to tackle such challenges. In addition, one can
use the idea of concept map and relevant computing tools. The
reasons are twofold. One of the reasons is the ease of
understanding—concept maps are able to express issues (for this
particular case a growing C–K map) in a human-friendly manner.
This can be verified from the concept maps shown in Figs. 1, 3–5
and 9. The other reason is the possibility of reuse and distribution
of C–K maps using Internet or other digital means. In this case,
there are many tools available for constructing concept maps (e.g.,
Cmap Tools [15,3], Hozo [13,16], and alike). Fig. 12. Internet based C–K mapping.
Even though concept map based tools are available for
building a C–K map, challenges remain for fostering creativity.
The framework shown by the concept map in Fig. 12 can be Internet. The goal, however, is to come up with concepts other
used to search the Internet while constructing a C–K map. In than the given one (i.e., other than C1). For example, consider the
fact, the authors have used this framework to build the C–K map phrase ‘‘propulsion engine other than fossil-fuel-based engine.’’
shown in Fig. 9. While doing this, the authors have come across Unfortunately, this phrase provides information of fossil-fuel-
some difficulties. These difficulties are described, as follows: based engines and engines that use renewable energy. The phrase
was rephrased to ‘‘propulsion engine other than fossil-fuel-based
Keywords: The wording and structure of keywords are two engine for Mars Exploration’’ by adding the phrase ‘‘Mars
important issues. To search creative concepts and underlying Exploration.’’ This phrase also produced similar result. This
knowledge, one can use a keyword (a phrase) consisting of means the goal (getting concepts other than C1) was not fulfilled.
physical phenomenon and object. For example, consider the One of the reasons of this undesirable result is perhaps the
phrase ‘‘combustion of fuel and Mars.’’ Here, ‘‘combustion of fuel’’ affinity of the exiting search engine toward commonly used
is the physical phenomenon and ‘‘Mars’’ is the object. The first phrase (for this case ‘‘fossil-fuel’’). Thus, searching for creative
two hits were http://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.- concepts using Internet needs a great deal of research in the near
gov/20040161238_2004146976.pdf and http://www.grc.nasa.- future wherein novelty of information should be empathized
gov/WWW/combustion/zPropellants.htm. Both refer to similar more that the relevance of information.
concept Z–CO2 based propulsion engine for Mars exploration (Z Useful/less useful websites: As illustrated in Fig. 12, the websites
means Al, Mg, etc.). Since enough work has been done on Z–CO2- found after the search should be classified into two categories:
based propulsion engine, it was possible to acquire knowledge Useful Websites and Less Useful Websites. Due to lack of time or
(C–K map in Fig. 9). If there were no works done on Z–CO2-based any other practical reasons, a user mostly likely take into
propulsion engine, the search fails to produce any result. In this account the content of the first few hits (websites) out of many
case, a user might use an existing concept (e.g., C1 = fossil-fuel- hits. This strategy was used while constructing the C–K map
based propulsion engine) to formulate a phrase for searching the shown in Fig. 9—only first 10 hits were investigated. This is
G Model
perhaps an ad hoc procedure. A rather systematic procedure is Appendix A. Calculating information content under epistemic
needed for distinguishing Useful Websites from Less Useful uncertainty
Websites while constructing a C–K map using Internet. This
issue also remains open for further research. Calculation procedure of information content of a design under
Knowledge validation: The trustworthiness of knowledge found abundance or lack of knowledge has been investigated by Ullah
in a website should be confirmed by the knowledge found in [24,25], Ullah et al. [28], and Ullah and Harib [27]. This appendix
another website and both websites should be independent from uses similar procedure but customized for two-proposition cases
each other. For example, consider the C–K map shown in Fig. 9. (cases in Tables 1–3).
The source of knowledge regarding Mg–CO2-(or Y–CO2) based Consider two propositions, Q1 and Q2, and a set of linguistic
engine is the research results of Shafirovich et al. [19,20] and it truth values T1, . . ., T5. The linguistic truth values can be
is confirmed by the work of Foote and Litchford [6]. Therefore, if expressed by membership functions of fuzzy numbers, mTi:
the search engine hits only few websites, wherein the contents [0, 1] ! [0, 1], i = 1, . . ., 5. The linguistic truth values used in this
of the hits are not contradictory, then it would be convenient for study are as follows: T1 = ‘‘mostly false,’’ T2 = ‘‘perhaps false,’’
a user to produce a trustworthy C–K map. At present, there is no T3 = ‘‘not sure,’’ T4 = ‘‘perhaps true,’’ and T5 = ‘‘mostly true.’
guarantee that a search will end up with websites wherein the These truth values are defined by the membership functions, as
contents are redundant and sources of the websites are follows:
independent from each other. However, peer-reviewed articles (
0:3 x
and resources can be considered validated information. Thus, mmostly false ðxÞ ¼ 0:3 0 x 2 ½0; 0:3 (A.1)
contents of open digital library (e.g., Universal Digital Library 0 otherwise
(UDL) (http://www.ulib.org)), domain knowledge digital library
8
(e.g., Kinematic Models for Design: Digital Library (http:// > x0
>
> x 2 ½0; 0:3
knoddl.library.cornell.edu)), digital library of scholarly articles < 0:3 0
(e.g., Scholarly and Academic Information Navigator (http:// mperhaps false ðxÞ ¼ 0:5 x (A.2)
>
> x 2 ½0:3; 0:5
ci.nii.ac.jp)), and alike can be used validated sources of >
: 0:5 0:3
0 otherwise
knowledge while inducing a creative concept. This issue also
needs further investigations. 8
> x 0:3
>
> x 2 ½0:3; 0:5
< 0:5 0:3
5. Concluding remarks mnot sure ðxÞ ¼ 0:7 x (A.3)
>
> x 2 ½0:5; 0:7
>
: 0:7 0:5
(a) The cognitive process involved in C–K theory is more complex 0 otherwise
than classical abduction. It is rather a motivation driven 8
> x 0:5
process. >
> x 2 ½0:5; 0:7
< 0:7 0:5
(b) Information content of design from the sense of epistemic mperhaps true ðxÞ ¼ 1x (A.4)
uncertainty should be maximized to remain creative. >
> x 2 ½1; 0:7
> 1
: 0:7
(c) When new knowledge is available, the information content 0 otherwise
should go down significantly. Otherwise, the new knowledge
(
does not add any value to the design process. x 0:7
mmostly true ðxÞ ¼ x 2 ½0:70; 1 (A.5)
(d) Human-friendly C–K map can be constructed by using 1 0:7
0 otherwise
concept map. If a C–K map in the form of concept map is
constructed by using the information available on the
Internet, new search engine and knowledge validation The membership functions defined in (A.1)–(A.5) are illustrated
mechanism will be needed. in Fig. A.1.
mostly false perhaps false not sure perhaps true mostly true
1
0.8
Membership Value
0.6
0.4
0.2
0
0 0.2 0.4 0.6 0.8 1
Truth Value
Fig. A.1. Membership functions of linguistic truth values.
G Model
The average numerical truth values of linguistic truth values Ti

denoted by TVTi, i = 1, . . ., 5 can be calculated as follows:
1
Z1
mTi ðxÞxdx
0 0.75
TVTi ¼ (A.6)
Z1
mTi ðxÞdx
RE
0.5
0
Thus, numerical truth values (TVTi) of linguistic truth values 0.25

‘‘mostly false’’ is equal to 0.1, ‘‘perhaps false’’ is equal to 0.27, ‘‘not
sure’’ is equal to 0.5, ‘‘perhaps true’’ is equal to 0.73, and ‘‘mostly
true’’ is equal to 0.9 (all values are rounded up to two digits). 0
For a given situation, linguistic truth values TQ1 and TQ2 can be 0 0.5 1
assigned to propositions Q1 and Q2, respectively, wherein TQ1, TVR
TQ2 2 {Ti j i = 1, . . ., 5}. The numerical counterparts of TQ1 and TQ2 is
TVQ1 and TVQ2. Therefore, TVQ1, TVQ2 2 {TVTi}, i = 1, . . ., 5. The Fig. A.3. An illustration of function RE.
values of TVQ1 and TVQ2 can be used to calculate the Certainty
Entropy (or Compliance) denoted by CE and Requirement Entropy
(or Compliance) denoted by RE. The equations are as follows: Table 1. Thus, the numerical truth values of Q1 is TVQ1 = 0.1
P2 (mostly false) and of Q2 is TVQ2 = 0.73 (perhaps true). Moreover,
j¼1 IðTVQ j Þ the numerical truth value of requirement proposition (R) is
CE ¼
2 TVR = 0.1 (=TVQ1) because Q1 is equivalent to requirement
so that 8
> TV 0 (A.7) proposition (engine should be suitable for Mars atmosphere). In
< Qj TVQ j 2 ½0; 0:5
0:5 0 addition, for this case a = max(TVQ1, TVQ2) = max(0.1, 0.73) = 0.73
IðTVQ j Þ ¼
: 1 TVQ j otherwise
> and b = min(TVQ1, TVQ2) = min(0.1, 0.73) = 0.1.
1 0:5 As a result, I(TVQ1) = (0.1 0)/(0.5 0) = 0.2 (see Fig. A.2) and
8 I(TVQ2) = (1 0.73)/(1 0.5) = 0.54 (see Fig. A.2), CC = (0.2 + 0.54)/
<1
> TVR b 2 = 0.37 according to Eq. (A.7) and RE = 1 according to Eq. (A.8)
a TVR
RE ¼ TVR 2 ðb; a (see Fig. A.3). Note that CC = 0.37 implies that the suitableness of an
>
: ab
0 TV > a (A.8) ordinary engine (C1) is somewhat clearly known and RE = 1
so that implies that it (C1) is not at all suitable for Mars atmosphere.
a ¼ maxðTVQ j Þ b ¼ minðTVQ j Þ It is worth mentioning that CC = 0 means the knowledge is
absolutely complete, CC = 1 means the knowledge is absolutely
In Eq. (A.8), TVR is the truth value of the requirement incomplete, and CC is between 0 and 1 means knowledge is
proposition. partially complete or incomplete. On the other hand, RE = 0 means
The nature of the function I(TVQi) is illustrated in Fig. A.2. absolute fulfillment of requirement, RE = 1 means absolute un-
The function of RE is illustrated in Fig. A.3 for a = 0.73 and fulfillment of requirement, and RE is between 0 and 1 means
b = 0.1. partial fulfillment or un-fulfillment of requirement.
The calculation process of CE and RE is further explained by
using the settings of the first two-proposition case in Table 1, i.e., References
the case underlying propositions P11 and P12. As such, Q1 = P11
(=C1 is suitable for Mars atmosphere) and Q2 = P12 (=C1 is not [1] Altshuller, G.S., 1985, Creativity as an Exact Science, Gordon and Breach, New York
.
[2] Braha, D., Reich, Y., 2003, Topological Structures for Modelling Engineering
suitable for Mars atmosphere). The linguistic truth values of Q1 is
Design Processes, Research in Engineering Design, 14/4: 185–199.
TQ1 = ‘‘mostly false’’ and Q2 is TQ2 = ‘‘perhaps true,’’ as shown in [3] Canas, A., Novak, J.D., 2006, The Theory Underlying Concept Maps and How to
Construct Them, Technical Report IHMC Cmap Tools 2006-01, Florida Institute
for Human and Machine Cognition, Florida, USA, 2006.
[4] Chakrabarti, A., 2011, Motivation as a Major Direction for Design Creativity
Research, in Taura T, Nagai Y, (Eds.) Design Creativity 2010. Springer, New
York , pp. pp.49–56.
[5] Dubois, D., Foulloy, L., Mauris, G., Prade, H., 2004, Probability–Possibility
1
Transformations, Triangular Fuzzy Sets, and Probabilistic Inequalities, Reliable
Computing, 10:273–294.
[6] Foote, J.P., Litchford, R.J., 2007, Powdered Magnesium—Carbon Dioxide Rocket
0.75 Combustion Technology for In Situ Mar Propulsion, NASA Technical Publica-
tion, NASA/TP-2007-215077, available from https://www2.sti.nasa.gov.
I(TV Qi )
[7] Gero, J.S., 2000, Computational Models of Innovative and Creative Design
Processes, Technological Forecasting and Social Change, 64/2–3: 183–196.
0.5 [8] Hatchuel, A., Weil, B., 2003, Proceedings of the 14th International Conference
on Engineering Design Research for Practice (ICED’03) (19–21 August 2003,
Stockholm, Sweden), A New Approach of Innovative Design An Introduction to
0.25 C–K Theory, 109–124.
[9] Hatchuel, A., Weil, B., 2009, C–K design theory: an advanced formulation,
Research in Engineering Design, 19/4: 181–192.
[10] Hatchuel, A., Masson, P.L., Weil, B., 2011, Teaching innovative design reason-
0 ing: How concept–knowledge theory can help overcome fixation effects,
0 0.5 1 Artificial Intelligence for Engineering Design Analysis and Manufacturing,
25/1: 77–92.
TVQi [11] Kazakci, A.O., Tsoukias, A., 2005, Extending the C–K Design Theory: A Theo-
retical Background for Personal Design Assistants, Journal of Engineering
Fig. A.2. Information content of a truth value. Design, 16/4: 399–411.
G Model
[12] Kazakci, A.O., Hatchuel, A., Weil, B., 2008, Proceedings of the International [22] Taura T, Nagai Y, (Eds.) (2011), Design Creativity 2010. Springer,
Design Conference Design 2008 (Dubrovnik: Croatia, May 19–22, 2008), A New York .
Model of CK Design Theory Based on Term Logic A Formal Background for a [23] Tomiyama, T., Gu, P., Jin, Y., Lutters, D., Kind, C., Kimura, F., 2009, Design
Class of Design Assistants, . Methodologies: Industrial and Educational Applications, CIRP Annals –
[13] Mizoguchi, R., Sunagawa, E., Kozaki, K., Kitamura, Y., 2007, A Model of Roles Manufacturing Technology, 58/2: 543–565.
Within an Ontology Development Tool: Hozo, Journal of Applied Ontology, 2/ [24] Ullah, A.M.M.S., 2005, A Fuzzy Decision Model for Conceptual Design, Systems
2: 159–179. Engineering, 8/4: 296–308.
[14] Novak, J.D., Gowin, D.B., 1984, Learning How to Learn, Cambridge University [25] Ullah, A.M.M.S., 2005, Handling Design Perceptions: An Axiomatic Design
Press, New York and Cambridge, UK. Perspective, Research in Engineering Design, 16/3: 109–117.
[15] Novak, J.D., Canas, A., 2006, The Origins of the Concept Mapping Tool and the [26] Ullah, A.M.M.S., 2008, Logical Interaction Between Domain Knowledge and
Continuing Evolution of the Tool, Information Visualization, 5:175–184. Human Cognition in Design, International Journal of Manufacturing Technol-
[16] Ohta, M., Kozaki, K., Mizoguchi, R., 2011, An Extension of the Environment for ogy and Management, 14/1–2: 215–227.
Building/Using Ontologies Hozo toward Practical Ontology Engineering—Fo- [27] Ullah, A.M.M.S., Harib, K.H., 2008, An Intelligent Method for Selecting Opti-
cused on Practical Issues, Journal of Japan Society of Artificial Intelligence, 26/ mal Materials and its Application, Advanced Engineering Informatics, 22/4:
2: 403–418. (in Japanese). 473–483.
[17] Reich, Y., Hatchuel, A., Shai, O., Subramaniam, E., 2010. A Theoretical Analysis [28] Ullah, A.M.M.S., Harib, K.H., Al-Awar, A., 2007. Minimizing Information
of Creativity Methods in Engineering Design: Casting and Improving ASIT Content of a Design Using Compliance Analysis, SAE Technical Paper
Within C–K theory, Journal of Engineering Design, in press. 2007-01-1209.
[18] Schurz, G., 2008, Patterns of Abduction, Synthese, 164/2: 201–234. [29] Yoshikawa, H., 1981, General Theory of Design Process, Journal of the Japan
[19] Shafirovich, E.Y., Goldshleger, U.I., 1992, Combustion of Magnesium Particles Society of Precision Engineering, 47/4: 405–410. (in Japanese).
in CO2/CO Mixtures, Combustion Science and Technology, 84/1: 33–43. [30] Zeng, Y., Cheng, G.D., 1991, On the Logic of Design, Design Studies, 12/3:
[20] Shafirovich, E.Y., Shiryaev, A.A., Goldshleger, U.I., 1993, Magnesium and Car- 137–141.
bon Dioxide: A Rocket Propellant for Mars Missions, Journal of Propulsion and [31] Zadeh, L.A., 1978, Fuzzy Sets as a Basis for a Theory of Possibility, Fuzzy Sets
Power, 9/2: 197–203. and Systems, 1:3–28.
[21] Suh, N.P., 1998, Axiomatic Design Theory for Systems, Research in Engineering [32] Galle, P., 2009, The ontology of Gero’s FBS model of designing, Design Studies,
Design, 10/4: 189–209. 30/4: 321–339.

Cirp Journal of Manufacturing Science and Technology: A.M.M. Sharif Ullah, Md. Mamunur Rashid, Jun'ichi Tamaki

Caricato da

Informazioni sul documento

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

Cirp Journal of Manufacturing Science and Technology: A.M.M. Sharif Ullah, Md. Mamunur Rashid, Jun'ichi Tamaki

Caricato da

Copyright:

Formati disponibili

G Model

CIRPJ-183; No. of Pages 12

CIRP Journal of Manufacturing Science and Technology xxx (2011) xxx–xxx

Contents lists available at SciVerse ScienceDirect

CIRP Journal of Manufacturing Science and Technology

On some unique features of C–K theory of design

1. Introduction co-creation of new knowledge) has been a subject of research. Some

1755-5817/$ – see front matter ß 2011 CIRP.

Fig. 1. An illustration of C–K theory of design.

Fig. 2. Implication of abduction.

the typical nature of continuing a design process by using classical

Fig. 4. Concept map of C–K theory based design process.

Fig. 5. A concept map of creative concept based on C–K theory.

Propositions Truth values Requirement

Fig. 6. Information content of C1. Fig. 7. Information content of C2.

false,’’ based on Mars atmosphere (K2). On the other hand, both

3.3. Effectiveness of new knowledge

Lack of knowledge (K4 = {1}) actually results a very high

Propositions Truth values Requirement

Fig. 9. C–K map when K4 transforms into K0 4.

Propositions Truth values Requirement

P1 C2 is acceptable in terms of Speciﬁc Impulse Perhaps true 0.73 An engine should be

Fig. 11. Unique characteristic of C–K theory based design process.

how to assist designers building a C–K map. Once a C–K map is

Fig. A.1. Membership functions of linguistic truth values.

The average numerical truth values of linguistic truth values Ti

Thus, numerical truth values (TVTi) of linguistic truth values 0.25

Potrebbero piacerti anche