A method for evaluating creativity in Linkography

A method for evaluating creativity in Linkography Shyan-Bin Chou Short biography of the author(s) Dr. Shyan-Bin Chou received B.S. degree of Civil and Hydraulic Eng. from Chung- Yuan Christian University,
of 6
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
A method for evaluating creativity in Linkography Shyan-Bin Chou Short biography of the author(s) Dr. Shyan-Bin Chou received B.S. degree of Civil and Hydraulic Eng. from Chung- Yuan Christian University, Taiwan, R.O.C. in He then received M.S. of Computer and Information Science from New Jersey Institute of Technology, U.S.A in And on 2002, he received PhD of Information Eng. from Chung-Yuan Christian University, Taiwan, R.O.C. His research interest includes: collaborative design, computer animation and digital art creation. He is currently director and associate professor of Graduate Institute of Design, National Taiwan Normal University, Taiwan, R.O.C. E_mail: Keywords: Linkography, entropy, collaborative design, creativity, pattern-matching algorithm. Research paper Introduction Effective communication is an important factor to the success of collaborative design. Individual thought can be shared in a brainstorming manner and result in design-thought inspiring. Although communication covers from social psychology, information theory, linguistics, drawing and gesturing etc. most scholars agree that design by face-to-face (closely coupled) mode is superior to media-involved (loosely coupled) one.(jeff WT Kan, John S Gero, 2005a) As team design activity becomes primary source for idea generation. Problems of how to record the transitional process of collaborative work and evaluate the effectiveness of outcome become imperative issues. Goldschmid (Goldschmidt, 1990) proposed to use Linkography to assess the design productivity. In her method, designers protocol which is derived either via video data, interview or discussions is represented by a set of moves and links between moves. A link connected two moves if they are associated whereas a node indicates the unconnected relation between any two moves. The design process can then be graphically seen as a combination of sequentially listed moves and links among them. Figure 1 shows terms used in Linkography. Conceptually, her idea provided a mechanism for tracing abstract design procedures which are basically the collection of participants inter-opinions. move node link layer Figure 1. terms used in Linkography Problems and related issues Some researchers further extended Linkography, for example: Van der Lugh (Remko van der Lugh, 2003) extended Linkography into three types according to link direction alternation and concluded that a well-integrated creative process has a large network of links, a low level of self-links and a balance of link types. Kan and Gero (Jeff WT Kan, John S Gero, 2004) distinguished links of Linkography into two types: social and design. Social links are neglected and design links are classified by three threshing criteria: Global/Local/in details, function/behavior/structure and Process/Analysis/Evaluate. As stated earlier, Linkography simply documents the on-going design practice in a faithful manner. The follow up task is the interpretation of Linkography. Two interesting questions that researchers seek to answer are: how to identify creativity of a design by Linkography and what are the critical moves in Linkography? To solve the first problem, Goldschmid defined index as the ratio of the number of links to the number of moves and use it as a benchmark for design productivity. It is intrinsic and straightforward; the idea behinds it is the richer the responses, the higher possibility for productive design is expected. Essentially, it is convinced for it reflects that the more designer involvement the better design turnout can be expected. But from the idea generation perspective, a dense links merely implies high participations and are not requirements for the generation of creativity. Kan and Gero (Jeff WT Kan, John S Gero, 2005c) considered the Linkography from the statistic viewpoint. They removed the linking lines leaving the nodes and treated each link as a point in 2D dimension. Then they analyzed the scattering distribution of links by calculating the means and standard deviations of X axes. The linking length is determined by maximum of Y and mean value reveals the depth of discussion. Kan and Gero (Jeff WT Kan, John S Gero, 2005a) suggested adopting information theory for measuring creativity in Linkography. The basic concept is the higher entropy reflects a richer idea generation process in the sense that degree of uncertainty has similar meaning to the design creativity. Entropy s method is objective and effective but the demerit is it measures the degree of uncertainty in a global manner as it calculates the entropy layer by layer under the probability model. In other word, the distribution condition within each layer is not taken into consideration. It results in two quite diverse links distributions would come out the same entropy which is apparently unreasonable. Proposed method In order to combat the above problem, the author proposes to employ patternmatching algorithm for determining the frequency of pattern occurrence. A pattern in this paper is defined as a set of continuous nodes/links of each layer in Linkography. The basic concept of this approach is the frequency of pattern appearance implies the degree of response repetition which is opposite to the degree of uncertainty and the degree of uncertainty is assumed to be similar to creativity in essence. We then take the frequency of pattern appearance as a factor for decreasing entropy. Each layer s entropy is adjusted by multiplying this factor before summarizing together. For clarity, we take the Figure 2 as an example: the entropies calculated by (Jeff WT Kan, John S Gero, 2005a) in layer 1-6 of two Linkographys are identical although node distributions are totally poles apart. With proposed method, entropy in layer 5 of left and layer 4 of right Linkography should be adjusted as pattern re-appearances exist. layer 6 layer 1 Figure 2. two Linkographys having same entropy with Jeff WT Kan and John S Gero s method To adopt pattern-matching algorithm, the simplest way is to apply comparison for each pattern in a brute-force manner which has time complexity of O(mn) in worst case and O(m) in best case; given a pattern of m characters to be found in a string with length of n characters. There exist some better pattern-matching algorithms, for instance: Knuth-Morris-Pratt (KMP) algorithm. KMP method benefits most if a set of characters are self-repetitive. In the proposed method, a link (L) is defined as a connection between two moves. If two moves are not associated; we then mark the intersection as blank(b). Given n moves, Linkography is treated as a composition of layer 1 to layer n-1. Each layer in Linkography is viewed as a string composed of set of character which either is L or B. Supposed m characters located in a specific layer, the proposed method is seeking to find the repeat times of a certain pattern. The pattern is defined as CiCi+1 Cj where i=1, j=2 to m-1. The repeat times R1, R2,...Rm-2 is summed up to S. The entropy (E) in this layer is then modified to (1-S/K)*E before adding to the entire entropy of Linkography, where K is maximum possible repeated patterns and equals to (m-2)=(m-1)(m-2)/2. Case study The Figure 3 is taken as example for explanation of the proposed method. LBLLBL Figure 3. a Linkography with layer 6 being labeled The Linkography shown above has 4 links in layer 6. The entropy method proposed by Gero is calculated as followed: Total amount of nodes is 7*(7-1)/2=21. Links shown in this layer is 4. Entropy in this layer is (-4/21)* log (4/21) 0.456 As pattern LB repeats once and LBL repeats as well, S= R1+ R2=1+1=2 and K=(6-1)*(6-2)/2=10, the proposed method yields entropy in this layer to (1-S/K)*E =(1-2/10)*0.456= Similarly, entropy in layer 5 is (-5/21)* log (5/21) 0.493, the modified entropy in layer 5 is (1-2/6)*E= The entropies in layer 3 and 4 are not altered as no pattern repetition occurs. Conclusion and discussion A method of modified entropy calculation for Linkography has been outlined. The appliance of entropy for Linkography intends to measure the creativity by possibility of links existences. Entropy for each layer is calculated by number of links followed by summing all layers entropy. It mainly considers the degree of randomization for links distribution in global view, however, the extent of irregular appearance of links within layer is not well measured. The author proposes to adopt pattern-matching algorithm to examine the existence of repetitive pattern inside a layer. A pattern here is defined as a series of characters composed by blanks (B) and/or links (L). The number of characters in pattern starts from 2 to m-1; given the number of nodes in layer is m. The accumulated number of repeated pattern is used as a factor for adjusting entropy for this layer. The proposed method conceptually meets the entropy method for determining the extent of uncertainty and irregularity. Distribution of local links (within layer) is taken into considered. This makes original entropy method more reasonable. However, the preliminary study treats found pattern owe equal weight, this may need further practical protocol analysis to justify. Furthermore, the current patternmatching method starts from two left characters and ends up at rightmost m-1 characters, this strategy is not theory-supportive, whether or not central-bases patternmatching would generate realistic outcome needs more experiments to verity. Reference Goldschmidt, G (1990), Linkography: Assessing Designing Design Productivity, Paper presented to the Cyberbetics and System, Singapore Jeff WT Kan, John S Gero (2004), A Method to Analyse Team Design Activities, University of Sydney Jeff WT Kan, John S Gero (2005), Can Entropy Indicate the Richness of Idea Generation in Team Designing?, University of Sydney Jeff WT Kan, John S Gero (2005), Entropy Measurement of Linkography in Protocol. University of Sydney, Studying Designers Jeff WT Kan, John S Gero (2005), Design Behavior Measurement by Quantifying Linkography in Protocol Studies of Designing, University of Sydney, Human Behavior in Design Remko van der Lugh (2003), Relating the Quality of Idea Generation Process to the Quality of the Resulting Design Ideas, Paper presented to the International Conference on Engineering Design, Stockholm, Sweden, August
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks