Comparison of Decision Support Systems with DEMATEL-SAW and DEMATEL-TOPSIS in the Process of Journal Acceptance: Case study in The Postgraduate E-Journal of State University of Malang

. Concerns about scientific publications have been proliferating in Indonesia. Consequently, the number of published journals in e-journal has been rapidly increasing over the past few years. The growing trend of e-journal hence requires a decision support system in its application. The DSS will help the reviewers in determining the eligibility of an article in the journal’s verification process. Several DSS methods such as DEMATEL, SAW, and TOPSIS, among others, are proposed to provide an effective means in the process. This research aims to present a solid comparison of two combined methods, DEMATEL-SAW and DEMATEL-TOPSIS, as they overcome each method’s shortcomings, in determining the eligibility of an article. The eligibility criteria have been determined as a guide. The calculation results show that the DEMATEL-SAW has a relatively higher degree of accuracy compared to that of DEMATEL-TOPSIS when fewer criteria variables are included, whereas the DEMATEL-TOPSIS method has a higher degree of consistency when being utilized on a variable with more criteria.


INTRODUCTION
Concerns about scientific publications have been proliferating. For the past 50 years, the number of published science journals has been rapidly increasing as seen from numerous researches that utilized data from various literature database such as Science Citation Index (SCI) and Social Science Citation Index (SSCI). Simultaneously, academic publishing has been undergoing major changes as it steadily makes the transition from the print to the electronic format, and many academic journals, therefore, had been published on new channels such as the internet, conferences, and digital archives (Larsen, 2010). This has encouraged many universities, including the Graduate School of State University of Malang, to actively develop e-journal system.
The role of Decision Support System (DSS) is hence required in the ever-growing trend of ejournal. DSS will help the verification process of the reviewers and provide assistance in determining the eligibility of a journal. There have been numerous developments of DSS by applying MCDM (Multi-Criteria Decision Making) method like AHP (Analytical Hierarchy Process), SAW (Simple Additive Weighting), DEMATEL (Decision Making Trial and Evaluation Laboratory), TOPSIS (Technique for Order Preferences by Similarity to Ideal Solutions), Fuzzy, MAUT (Multi-Attribute Utility Theory), SMART (Simple Multi-Attribute Rating Technique). Among many others, DEMATEL, SAW, and TOPSIS are proven to be the most effective to be implemented in the journal recommendation submission system. DEMATEL is capable of analyzing the correlation between the components of each criterion (Aksakal, 2010), while SAW and TOPSIS are intuitive in the decisionmaking process with simple calculation (Velasquez, 2013).
The DEMATEL method is a methodology that can be employed to evaluate the factor effectiveness by narrowing down the criteria (Thor, 2013). The method describes the relation E as a direct-relation matrix (Chang, 2011) to visualize the causal relationships and the analysis of dominant criteria in a system; DEMATEL, however, requires another method like SAW and TOPSIS in order to stop receiving any recommendations from the journals (Tzeng, 2007).
The SAW method (Simple Additive Weighting), furthermore, is one of the most versatile and natural multi-criteria among other evaluation methods (Podvezko, 2011;Talebanpour, 2015). Huang et al. studied Multi-Attribute Utility Theory (MAUT) in which a decision can be taken through a single attribute (Abdullah, 2014). This method integrates value and weighting criteria into an estimated value called method criteria. The evaluation score is calculated for each alternative by multiplying the value of the attribute scale by the weight of the relative importance (Talebanpour, 2015). The SAW method requires the process of normalizing the decision matrix (X) to a scale that is comparable to all existing alternative ratings (Josaputri, 2016). The advantage of the SAW method is that the linear proportional transformation of the raw data results in the relative order of magnitude of the standard score that remains the same (Afshari, 2010).
On the other hand, Hwang and Yoon introduced TOPSIS in 1993 as a multi-criteria decisionmaking method that can rank a set of factors by identifying and weighing their interests, prioritizing them in a definite order (Khanjankhani, 2016). The method calculates the geometric distance between both alternatives and takes the result into account to get the most ideal alternative or the best result possible (Bulgurcu, 2012).
This research, therefore, presents a comparison between two combined approaches, DEMATEL-SAW and DEMATEL-TOPSIS, to provide a means in evaluating and determining the eligibility of a journal submitted in the e-journal of Postgraduate Study of State University of Malang.

METHODS
This section of the study presents the literature review related to the implementation of DEMATEL and SAW as well as the selection of the criteria that later will be used to evaluate the eligibility of the articles. The following is the criteria used as shown in Table 1.

Significance
How important is the work reported; does it involve or synthesize ideas, methods, approaches from multiple disciplines; does it have interesting implications for multiple disciplines Originality Is this a new issue; does the paper point out differences from related research Quality Is the paper technically sound; how are its claims backed up; Does it carefully evaluate the strengths and limitations of its contribution Clarity Is the paper clearly written, does it motivate the research; does it describe clearly the methods employed (e.g., experimental procedures, algorithms, analytical tools); Are the results, if any, described and evaluated thoroughly Relevance Is the paper closely related to the theme; is the content interesting enough to a broad audience; From the abovementioned aspects, we have obtained the criteria required to get the weighting result. To collect the sample of data, the respondents were asked to complete a questionnaire, in which each criterion will determine the weight-taking of the data. The questionnaire can be seen in Table 2.  (Sumrit, 2013). Dematel is used to analyze and establish causal relationships that occur between evaluation criteria (Sumrit, 2013). This method uses a diagram and matrix to describe the causal relationships that occur between the dominant criteria (Chang, 2011).
Here are the steps in the DEMATEL method: 1. First, calculate the initial average matrix by score. Subsequently, the respondents are asked to show the level of direct influence of each factor or element we give to each factor or element j, which is denoted by "aij". We assume that scales 0, 1, 2, 3 and 4 represent ranges of "no effect" to "very high impact". Each respondent will produce a direct matrix, and the average matrix A is then derived through the mean of the same factor or element in the various direct matrix of the respondent. The average matrix A is represented in the following equation: 2. Second, calculate the normalized initial direct-influence matrix, . It can be obtained through normalizing the average matrix A. Specifically, the matrix X can be obtained through the equation (2) and (3), in which all the main diagonal elements are equal to zero.
3. Third, derive the direct / non-direct matrix directly. The continuous derivation of the non direct effect along the line of X. For instance, and , in which and/or with only one equal column or row. 1 total-influence matrix is registered as follows, where . Furthermore, this method presents the result of every sum and the number of columns of matrix T, in which ri shows the number of rows from the i row of matrix T and the direct and non-direct effect of the other i in other factors / elements. Likewise, cj shows the number of columns from the j column of matrix T as well as the nondirect effects of other factors / elements of other factors / criteria. In addition, when i is equal to j (or the sum of the columns and rows), (ri + ci) gives an index of power given and accepted. (ri + ci) indicates the key role of the factors. If (ri + ci) is positive, i will influence other factors; however, if it is negative, i will be influenced by other factors (Tzeng, 2007). 4. Fourth, set the threshold values and get the IRM. Setting the threshold value A, filtering out the minor effects represented by the matrix factor T, is necessary in order to isolate the structure of factor relations. Based on the matrix T, each matrix factor tij informs about how i influences j. Practically, if all information from a T matrix converts to IRM, the map would be too complex to show the information required for the decision-making process. To reduce the complexity of IRM, decision makers have to set the threshold value for the level of influence: that only the factor whose value of influence on the T matrix is higher than the threshold value can be selected and converted to IRM. The threshold value can be decided through a brainstorming process of experts. IRM can be shown when the relative threshold and IRM values have been decided (Chou, 2016).

Simple Additive Weighting (SAW)
The SAW method, usually dubbed the simple weighting method, represents each criterion that is multiplied by a certain weight (Velasquez, 2013). The calculation integrates a particular criterion and weight value into a single unit, indicated by the method criteria (Sj). The result of Sj from the weighted normalization of each criterion is then calculated for j using the following equation (Podvezko, 2011): is the weigh of the i (∑ =1 = 1); is the result of normalization of i for j; i=1,…,m; j=1,…,n; m is the number of criteria used while n is the object being compared. TOPSIS TOPSIS (Technique for Order Preferences by Similarity to Ideal Solutions) is a method used to identify alternative solutions in order to approach positive ideal solutions and avoid negative ideal solutions (Thor, 2013). The positive ideal solution shows the maximum solutions that are determined from the combination of best values of the matrix calculations; meanwhile, a negative ideal solution shows a minimal solution gain, denoting the combined result of the worst value of the matrix calculation. The positive ideal solution is shown as + and the negative ideal solution is shown as − . The following is the calculations for finding both the positive and negative ideal solution (Thor, 2013)

RESULTS AND DISCUSSIONS Data Processing with the Dematel Method
We use the DEMATEL method to find out the weight and interrelationship between the criteria. The data used are sourced from the questionnaire. The matrix questionnaire is shown in Table 3. The next step is to find out the relationship matrix directly by normalizing the questionnaire matrix of the DEMATEL method. The direct link matrix is shown in Table 4. R is the sum of the columns, whereas D is the total of the direct-relationship matrix rows. The values of D and R will be used in determining the relationship between variables, grouping them into two groups: dispatcher and receiver. The dispatcher are the criteria that become the cause of other criteria, whereas the receiver becomes the result of other variables. The dispatcher and receiver groups are shown in Table 5. The significance, originality, and quality of an article belong to the dispatchers-these are the main criteria that will influence other criteria. On the other hand, the clarity and relevance are included in the receiver group, which in turns is influenced by those who belong to the dispatchers. Therefore, in reviewing journals, it is suggested that one prioritizes the significance, originality and quality as the main criteria

Data Processing with the Dematel Method
The data processing with the SAW method is used to accommodate the result of weights calculation of that of DEMATEL and proceed to the actual calculation, which is the recommendation of the journal's acceptance. The result of the DEMATEL data processing is shown in Table 6. The data processing using SAW method will result in values and ranks that will be used to recommend which journals are accepted (in this research, the value ranges from 1 to 3). Subsequently, it will be compared with the results of the experts in order to find the standard error and its accuracy. The standard error of the method is 11,97% and the accuracy is 88,03%.

Data Processing with the TOPSIS Method
The TOPSIS method is used to execute the weighting values that have been calculated using the DEMATEL method. The result of the TOPSIS data processing is shown in Table 7.

Comparison between Dematel-SAW and Dematel-TOPSIS Method
The calculation result of the DEMATEL-SAW and the DEMATEL-TOPSIS will be compared to see which method is the most effective in the decision-making process in the journal's acceptance. The result of the comparison is shown in table 8.
The calculation using the TOPSIS method will result in the values and ranks that contribute in the journal's acceptance. In determining the accuracy and the standard error, it is suggested that the result is compared to that of the experts. The standard error of the TOPSIS method is 34,82% and the accuracy is 65,13%. The result of the calculation shows that the accuracy level of the DEMATEL-SAW is relatively higher compared to that of the DEMATEL-TOPSIS, whereas the result of the DEMATEL-TOPSIS method shows a more consistent accuracy when being implemented on variables with numerous criteria. Hence, it is concluded that DEMATEL-SAW is slightly better as it has a lower rate of standard error.

CONCLUSION
The result of the DEMATEL method in the process of e-journal selection suggests that the reviewer should be more concerned with several main criteria such as significance, originality and quality, for they affect other criteria like clarity and relevance. In this research, the result also shows that the DEMATEL-SAW method befits the data processing if it includes fewer criteria as its high accuracy but inconsistent when the criteria used are many.