Decision making software architecture; the visualization and data mining assisted approach

www.iji-cs.org

International Journal of Information and Computer Science (IJICS) Volume 3 Issue 1, January 2014 doi: 10.14355/ijics.2014.0301.03

Decision-making Software Architecture; the Visualization and Data Mining Assisted Approach A. Mosavi University of Debrecen Egyetem ter 1, Debrecen 4031, Hungary a.mosavi@math.unideb.hu Abstract

convenient way [9].

Multiple Criteria Decision Making (MCDM) is considered as a combined task of optimization and decision-making. Yet in solving real-life MCDM problems often most attention is paid to finding the complete Pareto-optimal set of the associated multiobjective optimization (MOO) problem and less to decision-making. In this paper, along with the presentation of two case studies, an interactive procedure has been suggested which involves the decision-maker (DM) in the optimization process helping to choose a single solution at the end. Our proposed method on the basis of reactive search optimization (RSO) algorithm and its novel software architecture packages of LION solver and Grapheur are capable of handling the big data often associated with MCDM problems. The two considered case studies are; firstly, a series of MCDM problems in construction workers’ management and secondly, the well-known MOO and decision-making problem of welded beam design.

The efficient MOO algorithms facilitate the DMs to consider multiple and conflicting goals of a MCDM problem simultaneously. Some examples of such algorithms and potential applications could be found in [3, 4, 5, 6, 13]. Within the known approaches to solving complicated MCDM problems there are different ideologies and considerations in which any decision-making task would find a fine balance among them e.g. [85, 86, 87, 88, 98].

Keywords Interactive Multiple Criteria Decision Making; Reactive Search Optimization; Multiobjective Optimization

Introduction The task of MCDM is divided into two parts: (1) an optimization procedure to discover conflicting design trad-offs and (2) a decision-making process to choose a single preferred solution among them. Although both processes of optimization and decision-making are considered as two joint tasks, yet they are often treated as a couple of independent activities [1, 2]. For instance evolutionary multiobjective optimization (EMO) algorithms [14, 15] have mostly concentrated on the optimization aspects, developing efficient methodologies of finding a set of Pareto-optimal solutions. However finding a set of trade-off optimal solutions is just half the process of optimal design. This has been the reason why EMO researchers have been trying to find ways to efficiently integrate both optimization and decision making tasks in a 12

In MCDM algorithms e.g. [10, 11, 12, 18, 66, 68, 69, 71, 73, 75, 78, 82] often the single optimal solution is chosen by collecting the DM’s preferences where MOO and decision making tasks are combined to obtain a point by point search approach. In addition, in MOO and decision-making, the final obtained solutions must be as close to the true optimal solution as possible and the solution must satisfy the preference information.

Towards such a task, an interactive tool to consider decision preferences is essential. This fact has motivated novel researches to properly figure out the important task of integration between MOO and MCDM [9, 22, 23, 14]. Naturally in MOO, interactions with the DM can come either during the optimization process, such as in the interactive evolutionary algorithms(EA)s optimization [15], or during the decision-making process [16, 37]. As a MOO task is not complete without the final decision-making activity and in this sense there exists a number of interactive MOO methods in the MCDM literature [23, 33, 39, 40].

MCDM and EMO; Integration There are two different ways by which EMO and MCDM methodologies can be combined together [14].

Either EMO followed by MCDM or MCDM integrated in an EMO. In the first way, an EMO algorithm is applied to find the Pareto front solutions. Afterward, a

International Journal of Information and Computer Science (IJICS) Volume 3 Issue 1, January 2014

single preferred solution is chosen from the obtained set by using a MCDM procedure. In this way, EMO application helps a DM to analyze different trade-off solutions before the final one is selected. However the DM has to go through analyzing many different solutions to be able to make the final decision.

Therefore the DM has to consider too many possible solutions.

Alternatively a MCDM procedure is integrated within an EMO to find the preferred Pareto front solutions where the search is concentrated on the important region of the Pareto front. This would let the optimization task evaluate the preferences of the DM interactively.

The approaches to interactive evolutionary algorithms are numerous. Takagi [17] reviewed various problem domains where an EA is carried out by the involvement of the DM. Additionaly a summary can be found in the text by Miettinen [18].

Some of the popular approaches are interactive surrogate worth trade-off method [19], the reference point method [20] and the NIMBUS approach [12]. All procedures require a DM to provide the preferences [14]. A search workflow is then used to find the optimum of the objective task. This procedure is repeated many times until the DM is satisfied with the obtained final solution. For instance in [21, 22] an EMO procedure is applied to a complicated design problem and then a interactive methodology is employed to choose a single solution. In [23], EMO is combined with MCDM procedures, and an interactive procedure is suggested where the EMO methodologies are combined with a certain and efficient MCDM technique. The work later in [23] is extended by involving more MCDM tools and integrations with further software packages such as MATLAB, to provide better working on more real-life case study [14]. In [23] unlike the classical interactive methods in [12], a good estimation of the Pareto-optimal frontier is created, in which helps to concentrate on a particular region. From this work, one could conclude that when a approach is best suited for one problem it may be inadequate in another problem. As the result it is worth mentioning that while developing MCDM and EMO integration, a successful procedure could include more than one optimization and decision-making tool in it so that any number of optimization and decisionmaking tool may be combined to build an effective problem solving procedure [24]. The researches reviewed above, have motivated other EMO and MCDM researches, including this article, to develop such integration schemes further.

www.iji-cs.org

Drowbacks to Solving MOO Problems Utilizing EAs The general form of a MOO problem, according to [7, 8], can be stated as; minimization of đ??&#x;đ??&#x;(đ?’™đ?’™) = {đ?‘“đ?‘“đ?&#x;?đ?&#x;? (đ?’™đ?’™), ‌ , đ?‘“đ?‘“đ?’Žđ?’Ž (đ?’™đ?’™)} , subjected to đ?’™đ?’™ ∈ Ί, where đ??ąđ??ą ∈â„?n is a vector of đ?‘›đ?‘› decision variables; đ??ąđ??ą ⊂â„?n is the feasible region and specified as a set of constraints on the decision variables; đ??&#x;đ??&#x; âˆś Ί →â„?m is made of đ?‘šđ?‘šobjective functions subjected to be minimization. Objective vectors are images of decision vectors written as đ??łđ??ł = đ??&#x;đ??&#x;(đ?’™đ?’™) = {đ?‘“đ?‘“đ?&#x;?đ?&#x;? (đ?’™đ?’™), ‌ , đ?‘“đ?‘“đ?’Žđ?’Ž (đ?’™đ?’™)}.Yet an objective vector is considered optimal if none of its components can be improved without worsening at least one of the others.

An objective vector đ??łđ??ł is said to dominateđ??łđ??ł ′ , denoted asđ??łđ??ł ≺ đ??łđ??ł ′ , if đ?‘§đ?‘§đ?‘˜đ?‘˜ ≤ đ?‘§đ?‘§đ?‘˜đ?‘˜â€˛ for allđ?‘˜đ?‘˜ and there has at least one â„Ž ďż˝ is Pareto optimal if there is no thatđ?‘§đ?‘§â„Ž ≤ đ?‘§đ?‘§â„Žâ€˛ . A point đ?’™đ?’™ ďż˝). The set of other đ?’™đ?’™ ∈ Ί such that đ??&#x;đ??&#x;(đ?’™đ?’™) dominates đ??&#x;đ??&#x;(đ?’™đ?’™ Pareto optimal points is called Pareto set (PS). And the corresponding set of Pareto optimal objective vectors is called Pareto front (PF).

Considering solving MOO problems EAs are among the most popular a posteriori methods to generate PS of a MOO problem. The EAs of MOO for solving MCDM problems have utilized for up to two decades. EAs are well suited to search for a set of PS to be forwarded to the DM. Yet EMOAs aim at building a set of points near the PF [79, 80, 81]. Currently, most EMOAs apply Pareto-based ranking schemes. Some of the most successful EMOAs [16, 24, 25] rely on Pareto dominance classification. However it becomes ineffective for increasing number of objectives. MOO of curve and surfaces [27, 37] would be a good example for such ineffective attempt and increasing complexity. The reviewed and applied approaches for solving the MOO of the curve and surfaces whether a priori or a posteriori, and in particular EA, involve plenty of various complications. The reason is that the proportion of PF in a set grows very rapidly with the dimension��. In fact the reality of applied DM has to consider plenty of priorities and drawbacks to both interactive and non-interactive approaches. Although the mathematical representative set of the DM model is often created presenting a human DM with numerous representative solutions on a multidimensional PF is complicated. This is because the typical DM cannot deal with more than a very limited number of information items at a time [63]. Therefore, an improved decision procedures should be developed according to human memory and one’s data processing capabilities. Moreover often DMs cannot formulate their objectives and preferences at the beginning. Instead they would rather learn on the job.

13

www.iji-cs.org

International Journal of Information and Computer Science (IJICS) Volume 3 Issue 1, January 2014

This has been already recognized in the MOO formulation, where a combination of the individual objectives into a single preference function is not executed. Considering the problem in [27, 37] the DM is not clear about the preference function. This uncertainty is even more increased when the objectives such as beauty are involved. This fact would inure lots of uncertainty and inconsistency.

Consequently interactive approaches try to overcome some of these difficulties by keeping the user in the loop of the optimization process and progressively focus on the most relevant areas of the PF directed by DM. This is done when the fitness function is replaced by a human user. However most DMs are typically more confident in judging and comparison than in explaining. They would rather answer simple questions and qualitative judgments to quantitative evaluations. In fact the identified number of questions that have to be asked are from the DM a crucial performance indicator of interactive methods. This would demand for selecting appropriate questions, for building approximated models which could reduce the bothering DM. The above facts, as mentioned in [39], and later in [36] demand a shift from building a set of PF, to the interactive construction of a sequence of solutions, so called brain-computer optimization [60], where the DM is the learning component in the optimization loop, a component characterized by limited rationality and advanced question-answering capabilities. This has been the reason for the systematic use of machine learning techniques for online learning schemes in optimization processes available in the software architectures of LION solver and Grapheur [39, 40, 59,60].

Brain-Computer Optimization (BCO) Approach to Stochastic Local Search As Battiti et al. [8, 39], also clearly stated that, the aim of stochastic local search is to find the minimum of the combinatorial optimization function đ?‘“đ?‘“ , on a set of discrete possible input values đ?‘‹đ?‘‹ . To effectively and interactively doing so the focus in [60] devoted to a local search, hinting at reactive search optimization (RSO)[26, 27, 28, 29, 36] with internal self-tuning mechanisms, and BCO i.e. a DM in the interactive problem-solving loop [67]. Accordingly in this context the basic problem-solving strategy would start from an initial tentative solution modifying the optimization function. According to [8] the local search starts from an acceptable configuration đ?‘‹đ?‘‹(0) and builds a search 14

trajectoryđ?‘‹đ?‘‹ (0) , . . . ,đ?‘‹đ?‘‹ (đ?‘Ąđ?‘Ą+1) , where đ?‘‹đ?‘‹ is the search space and đ?‘‹đ?‘‹ (đ?‘Ąđ?‘Ą) is the current solution at iteration đ?‘Ąđ?‘Ą, time. đ?‘ đ?‘ (đ?‘‹đ?‘‹ (đ?‘Ąđ?‘Ą) )then would be the neighborhood of point đ?‘‹đ?‘‹ (đ?‘Ąđ?‘Ą) ,obtained by applying a set of basic moves đ?œ‡đ?œ‡0 , đ?œ‡đ?œ‡1 , ‌ , đ?œ‡đ?œ‡đ?‘€đ?‘€ to the configuration of đ?‘ đ?‘ ďż˝đ?‘‹đ?‘‹ (đ?‘Ąđ?‘Ą) ďż˝ = { đ?‘‹đ?‘‹ ∈ đ?‘Ľđ?‘Ľ Suchthat đ?‘‹đ?‘‹ = đ?œ‡đ?œ‡đ?‘–đ?‘– ďż˝đ?‘‹đ?‘‹ (đ?‘Ąđ?‘Ą) ďż˝, đ?‘–đ?‘– = 0, . . . , đ?‘€đ?‘€} .If the search space is given by binary strings with a given lengthđ??żđ??ż: đ?‘‹đ?‘‹ = {0,1}đ??żđ??ż , the moves can be those changing the individual bits, and therefoređ??żđ??ż is equal to the string lengthđ?‘€đ?‘€.The accuracy of the achieved point is a point in the neighborhood with a lower value of đ?‘“đ?‘“ to be minimized. The search then would stop if the configuration is a local minimizer [7].

Y â&#x2020;?IMPROVING-NEIGHBOR (đ?&#x2018; đ?&#x2018; ďż˝đ?&#x2018;&#x2039;đ?&#x2018;&#x2039; (đ?&#x2018;Ąđ?&#x2018;Ą) ďż˝) đ?&#x2018;&#x152;đ?&#x2018;&#x152; if đ?&#x2018;&#x201C;đ?&#x2018;&#x201C;(đ?&#x2018;&#x152;đ?&#x2018;&#x152;) < đ?&#x2018;&#x2039;đ?&#x2018;&#x2039; (đ?&#x2018;Ąđ?&#x2018;Ą) đ?&#x2018;&#x2039;đ?&#x2018;&#x2039; (đ?&#x2018;Ąđ?&#x2018;Ą+1) = ďż˝ (đ?&#x2018;Ąđ?&#x2018;Ą) đ?&#x2018;&#x2039;đ?&#x2018;&#x2039; otherwise (search stops)

Here the local search works very effectively. This is because the improving-neighbor can identify and return an improving element in the neighborhood.

This is mainly because most combinatorial optimization problems have a very rich internal structure relating to the configuration đ?&#x2018;&#x2039;đ?&#x2018;&#x2039; and the đ?&#x2018;&#x201C;đ?&#x2018;&#x201C;value [8]. In the neighborhood the vector containing the partial derivatives is the gradient, and the change of đ?&#x2018;&#x201C;đ?&#x2018;&#x201C; after a small displacement is approximated by the scalar product between the gradient and the displacement [38].

Learning Component; the User in the Loop In problem-solving methods of stochastic local search proposed in [60] where the free parameters are tuned through a feedback loop, the user is considered as a crucial learning component in which different options are developed and tested until acceptable results are obtained. As explained in [7] by inserting the machine, learning the human intervention is decreased by transferring intelligent expertise into the algorithm itself. Yet in order to optimize the outcome setting of the parameters a simple loop is performed where the parameters in an intelligent manner are changed until a suitable solution is identified. Additionally, to operate efficiently, RSO uses memory and intelligence, to recognize ways to improve solutions in a directed and focused manner.

In the RSO approach of problem solving the braincomputer interaction is simplified. This is done via learning-optimizing process which is basically the insertion of the machine learning component into the solution algorithm. In fact, the strengths of RSO are associated to the brain characteristics i.e. learning from the past experience, learning on the job, rapid analysis

International Journal of Information and Computer Science (IJICS) Volume 3 Issue 1, January 2014

of alternatives, ability to cope with incomplete information, quick adaptation to new situations and events [7, 8]. Moreover the term of intelligent optimization in RSO refers to the online and offline schemes based on the use of memory, adaptation, incremental development of models, experimental algorithmics applied to optimization, intelligent tuning and design of heuristics. In this context with the aid of advanced visualization tools implemented within the software architecture packages [59] the integration of visualization and automated problem solving and optimization would be the center of attention.

RSO and Visualization Tools; an Effective Approach to MCDM Visualization is an effective approach in the operations research and mathematical programming applications to explore optimal solutions, and to summarize the results into an insight, instead of numbers [31, 32].

Fortunately, during past few years, there has been huge development in combinatorial optimization, machine learning, intelligent optimization, and RSO [7, 8, 33], which has moved the advanced visualization methods even further [32]. Previous work in the area of visualization for MCDM [32, 64, 65, 74] allows the DM to better formulate the multiple objective functions for large optimization runs. Alternatively in our research utilizing RSO and visualization [33], which advocates learning for optimizing, the algorithm selection, adaptation and integration are done in an automated way and the user is kept in the loop for subsequent refinements. Here one of the crucial issues in MCDM is to critically analyze a mass of tentative solutions assosiated with big data, which is visually mined to extract useful information [34, 35, 36].

In developing RSO in terms of learning capabilities there has been a progressive shift from the DM to the algorithm itself, through machine learning techniques [7, 90].

Concerning solving the MCDM problems, utilizing RSO, the final user is not distracted by technical details, instead concentration is on using his expertise and informed choice among the large number of possibilities. Algorithms with self-tuning capabilities like RSO make the interaction simpler for the final user.

And to doing so the novel approach of RSO is to integrate the machine learning techniques, artificial intelligence, reinforcement learning and active learning into search heuristics. According to the original literature [33] during a solving process the alternative solutions are tested through an online

www.iji-cs.org

feedback loop for the optimal parametersâ&#x20AC;&#x2122; tuning.

Therefore, the DM would deal with the diversity of the problems, stochasticity, and dynamicity more efficiently. Here are some case studies treated very promising by RSO. It is worth mentioning that RSO approach of learning on the job is contrasted with offline accurate parameter tuning [39, 40] which automatically tunes the parameter values of a stochastic local search algorithm.

Characteristics of the Proposed Approach During the process of solving the real-life problems exploring the search space, utilizing RSO, many alternative solutions are tested and as the result adequate patterns and regularities appear. While exploring, the human brain quickly learns and drives future decisions based on the previous observations and searching alternatives. For the reason of rapidly exploiting the most promising solutions, the online machine learning techniques are inserted into the optimization engine of RSO [8]. Furthermore, with the aid of inserted machine learning, a set of diverse, accurate and crucial alternatives is offered to the DM. The complete series of solutions are generated. After the exploration of the design space, making the crucial decisions, within the multiple existing criteria, fully depends on several factors and priorities which are not always easy to describe before starting the solution process. In this context, the feedbacks from the DM in the preliminary exploration phase can be incorporated so that a better tuning of the parameters takes the preferences into account [7]. Further relevant characteristics of RSO, according to [8], could be summarized as; learning on the job, rapid generation, and analysis of many alternatives, flexible decision support, diversity of solutions and anytime solutions.

Applications A number of complex optimization problems arising in widely different contexts and applications have been effectively treated by the general framework of RSO e.g. [72, 91, 92], including the real-life applications, computer science and operations research community combinatorial tasks, applications in the area of neural networks related to machine learning and continuous optimization tasks. This would include risk management, managing the big data of social networks, transportation, healthcare, marketing and ecommerce [59]. In addition, in the following some applications are summarized in real-life engineering application areas which are the main interests of this 15

www.iji-cs.org

International Journal of Information and Computer Science (IJICS) Volume 3 Issue 1, January 2014

research. In the area of electric power distribution, a series of real-life applications has been reported [42].

An open vehicle routing problem [43], as well as the pickup and delivery problem [44] both with the time and zoning constraints is modeled where the RSO methodology is applied to the distribution problem in a major metropolitan area. Alternative to solving the vehicle routing problem with backhauls, a heuristic approach based on a hybrid operation of reactive tabu search is proposed in [45]. By utilizing the RSO, the flexible job-shop scheduling [46], the plant location problem [47], the continuous flow-shop scheduling problem [48], adaptive self-tuning neurocontrol [49] and the real-time dispatch of trams [50] were effectively solved. Moreover, various applications of RSO focused on problems arising in telecommunication networks, internet and wireless in terms of optimal design, management and reliability improvements e.g. [51]. The multiple-choice multidimensional knapsack problem with applications to service level agreements and multimedia distribution is studied in [52]. In the military related applications, in optimal designing of an unmanned aerial vehicle routing system [53] and in finding the underwater vehicle trajectories [54], RSO worked wonder. The problem of active structural acoustic control [55] and visual representation of data through clustering [56] are also well treated. Additionally, the solution of the engineering roof truss design problem is discussed in [57]. An application of RSO for designing barrelled cylinders and domes of generalized elliptical profile is studied in [58]. Overall, a series of successful projects accomplished with the aid of RSO could be found in [25, 26, 27, 28, 29, 30]. Further applications of RSO are listed in [33, 40] and the stochastic local search book [41].

Software Architecture Packages for the Proposed Reactive and Interactive MCDM Grapheur and LION solver [7, 8, 39, 40, 59] are two implementions of RSO. The software implements a strong interface between a generic optimization algorithm and DM. While optimizing the systems produces different solutions, the DM is pursuing conflicting goals, and tradeoff policies represented on the multi-dimensional graphs [33, 40].During multidimensional graphs visualization in these software packages, it is possible to call user-specific routines associated with visualized items. This is regarded as the starting point for interactive optimization or problem solving attempts, where the user specifies a

16

routine to be called to get information about a specific solution. These implementations of RSO are based on a three-tier model, independent from the optimization algorithm, effective and flexible software architecture for integrating problem-solving and optimization schemes into the integrated engineering design processes and optimal design, modeling, and decisionmaking.

For solving problems with a high level of complexity, modeling the true nature of the problem is of importance and essential. For this reason, a considerable amount of efforts is made in modeling the MOO problems in Scilab which later will be integrated into optimizer package. Here, as an alternative to the previous approaches [22, 23, 24], the robust and interactive MOO algorithm of RSO is proposed in order to efficiently optimize all the design objectives at once in which couldnâ&#x20AC;&#x2122;t be completely considered in the previous attempts. In this framework, the quality of the design similar to the previous research workflows, is measured using a set of certain functions. Then an optimization algorithm is applied in order to optimize the function to improve the quality of the solution. Once the problem is modeled in scilab, it is integrated to the optimizer via advanced interfaces to the RSO algorithm and its brain-computer EMO implementations and visualization. In this framework, the application of learning and intelligent optimization and reactive business intelligence approaches in improving the process of such complex optimization problems is accomplished. Furthermore, the problem could be further treated by reducing the dimensionality and the dataset size, multi-dimensional scaling, clustering and visualization tools.

Study Case 1. Construction Workers In a construction project, a number of workers were surveyed with questionnaires and observations [29, 61, 62]. Each row of our data-set is a construction worker with the corresponding columns, characterized by a series of parameters i.e. ID and photo of each person, work time, looking for materials, looking for tools, specialization, moving, instruction, idleness and the other characteristics of the construction workers [28, 29, 30]. In fact, making critical decisions for the complicated and multiple criteria problems of construction projects in which huge amount of data, so called big data, is involved, is not a simple task to do as the problem has to be considered from different perspectives and criteria; for instance, the making

International Journal of Information and Computer Science (IJICS) Volume 3 Issue 1, January 2014

decision on how and with what rate the training programs should be organized, and what would be its further effects on team efficiencies, team spirit, and team perceptions of supervision. Here, the idea to solve the MCDM problems is to visually model and clarify the whole dimension of problems. As the main result, the 7D plots and the option of sweeping through data are to be considered for the cases. The achieved hidden information through visualization tools would enhance further decisions. In order to better manage the data collected and make the most of our data-sets we utilized the advanced interactive visualization tools provided by Grapheur software architecture where the user is connected to the software through automated and intelligent selftuning methods on the basis of visualization. With the aid of provided data mining and visualization, the useful and hidden information is achieved, which would enhance the process of solving MCDM problems of our cases [66, 67, 83, 84, 87]. After clarifying the dimension of the problem and finding out the relation between involved parameters and objectives, the effective decisions are easier made.

www.iji-cs.org

bubbles is proportional to the idleness of workers. In our similarity map of the graphical visualization, the gray level of the edges and the generated clusters provide valuable information to DM. In Fig. 1, the capability of the similarity map for an effective clustering of the workers into meaningful clusters is illustrated (Fig. 1 right).

The parallel filters (Fig. 1 left) are other useful tools for optimization. The usefulness of parallel filters in reducing the complexity from the process of decision making is evaluated. We start from the matrix of work time in a multidimensional space while aiming at filtering particular workers and examining their performance within a particular group e.g. those who have had maximum idleness characteristic.

Displaying the Precise Condition of Each Construction Worker

Supporting the Decisions on Workersâ&#x20AC;&#x2122; Skills

For the complete visualization of the condition of each construction worker over all parameters, the colored bubble chart is selected. In Fig. 2, (left), the colored bubble chart shows work time versus specialization for each worker. The color code and the size of the bubbles represent looking for material characteristic and the idleness status of the workers respectively.

With which rate and how, the workersâ&#x20AC;&#x2122; level of skills should grow in order to maintain their performance with regard to team perceptions of supervision. In order to study a part of this problem, the similarity map and the parallel filters to optimize the idleness characteristic of the workers have been taken into consideration. The related multidimensional plot of the networks is created based on the collected data from the workers. The color code represents the specialization of the workers and the size of the

Additionally, the shape of the bubbles displays the looking for tools characteristic of the workers. In this graph, we have found clustering tool very useful for a deep understanding of the different groups of workers.

In this case, workers could be grouped according to the given characteristics. After grouping, one prototype case for each cluster is visualized which is indeed a very effective way of compressing the information and concentrating on a relevant subset of possibilities.

FIG. 1PARALLEL FILTER (LEFT SIDE) AND SIMILARITY MAP (RIGHT SIDE)

17

www.iji-cs.org

International Journal of Information and Computer Science (IJICS) Volume 3 Issue 1, January 2014

FIG. 2 BUBBLE CHART (LEFT) AND SWEEPING THROUGH DATA IN THE BUBBLE CHART (RIGHT)

Sweeping Though Different Characteristics of Workers; Tracking and Examining the Problem with the Aid of Animated Graphs In the previous graph (Fig. 2. left) the relations between work time, specialization, idleness status, looking for materials, and looking for tools characteristics of the construction workers were visualized. Moreover, sweeping though data and studying the generated animations on sweeping is an effective tool for further visualization along with advancing a particular objective. For instance, in our next visualization experience, the previous graph is reconsidered by sweeping though looking material, idleness and skill level as the time advances (illustrated in Fig. 2 right).

Analyzing a Particular Cluster of Workers and Their Characteristics; Sweeping through Skill Level and Team Perception of Supervision

In the new created bubble graph, Fig. 3.(Right), the idleness and specialization characteristic of a cluster of four workers are associated with the size and the color of the bubbles respectively.

Here by sweeping through team perception of supervision and the level of specialization of field workers in our building construction project, the achieved information from a limited cluster of workers can clarify the problem with more details in different scenarios. For instance, when the skill level of the workers and the team perception of supervision are monthly increased relatively by the rate of 10% and 5% within a year, the idleness characteristic is smoothly monitored. We can also play the resulted animation in smooth mode and track the past values (they appear in a lighter tone in the background of the plot), in order to focus on the changes which occur according to morning and afternoon working shifts.

FIG. 3 7D PLOT OF DATA (LEFT) AND SWEEPING THROUGH DATA (RIGHT)

18

International Journal of Information and Computer Science (IJICS) Volume 3 Issue 1, January 2014

Providing a Reliable Way to Find the Most Productive Workers With the aid of a 7D plot, the characteristics associated with the productivity come to our consideration within a single graph. In our case, the size, the color and the shape of the bubbles relatively display the specialization, the moving, and the following the instructions characteristic of the workers. Moreover, the blinking feature displays the idleness characteristic of the workers who have been idle less than 100 hours (Fig. 3. left).

Discussion Along with our case studies, the aspects of data mining, modeling, and visualization the data related to construction workers are considered. We made the most of RSO applications via newly implemented data mining and visualization tools of Grapheur.

Considering the ability of Grapheur, the interesting patterns are automatically extracted from our raw data-set via data mining tools. Additionally the advanced visual analytical interfaces are involved to support the DM interactively. With the further feature of Grapheur such as parallel filters and clustering tasks, the managers can solve MOO problems as it amends previous approaches. Furthermore, the animations of sweeping through data and advanced visualizations including 7D plots accomplish managers and enable them to screen the data at their consulting room making decision interactively. In one of our case studies, Grapheur provided a widespread view on how the through put of the whole project would be affected by the increasing workersâ&#x20AC;&#x2122; specialization and supervision. We swept through different characteristic of workers in order to examine the whole dimensions of the problem. For instance, it was assumed that the problem having high level of idleness within the workers might be solved by increasing the supervision and team perception of supervision. For this reason, workers are carefully clustered and analyzed with regard to their level of idleness and supervision. In this particular case, Grapheur has been a facile tool in modeling the problem with the aid of a 7D plot. Once a 7D plot is created the problem could be visually analyzed from seven different perspectives simultaneously. In other words, a convenient way to concentrate on our objectives and further decision making is provided by simply observing the size, the color, the shape, and the blinking of the bubbles.

Moreover, utilizing further visualization options such as similarity maps, parallel filters, and clustering

www.iji-cs.org

would support making a confident decision.

Case Study 2.Welded Beam Design The problem of welded beam design [24] is a wellknown example of some complex designs issues arising in structural engineering, dealing with designing the form of steel beams and with connecting them to form complex structures. This case study has been used by many experts as a benchmark problem of single and also multiobjective design optimization.

The problem of designing an optimal welded beam consists of dimensioning a welded steel beam and the welding length in order to minimize the cost subjected to bending stress, constraints on shear stress, the buckling load on the bar, the end the deflection of the beam, and side constraints. There are four design variables i.e. h, l, t, b shown in the Fig. 4. Structural analysis of the welded beam leads to two nonlinear objective functions subjected to five nonlinear and two linear inequality constraints. The objectives are the fabrication cost and the end deflection of the beam. In our case, the aim is to reduce fabrication cost without causing a higher deflection. Decision-making on the preferred solution among the Pareto-optimal set requires the intelligent participation of the designer, to identify the trade-off between cost and deflection.

FIG.4 THE WELDED BEAM OPTIMAL DESIGN PROBLEM.

As shown in the above figure, the beam is welded on another beam carrying a certain load P. The problem is well studied as a single objective optimization problem [24], but we have transformed the original single objective problem into a two objective problem for more flexible design. In the original study, the fabrication cost (đ?&#x2018;&#x201C;đ?&#x2018;&#x201C;1 (x)) of the joint is minimized with four nonlinear constraints related to normal stress, shear stress, buckling limitations and a geometry constraint. With the following formulation, we have introduced one more objective i.e. minimization of the

19

www.iji-cs.org

International Journal of Information and Computer Science (IJICS) Volume 3 Issue 1, January 2014

end deflection (đ?&#x203A;żđ?&#x203A;ż(đ?&#x2018;Ľđ?&#x2018;Ľ)) of the structure. The problem has four decision variable presented in the optimization formulation, i.e. thickness of the beam b, width of the beam t, length of weld l, and weld thickness h. The overhang portion of the beam has a length of 14 in and F Âź 6; 000 lb force is applied at the end of the beam.

The mathematical formulation of the problem is given as; Minimizeđ?&#x2018;&#x201C;đ?&#x2018;&#x201C;1 (x) = 1.104711â&#x201E;&#x17D;2 đ?&#x2018;&#x2122;đ?&#x2018;&#x2122; + 0.04811đ?&#x2018;Ąđ?&#x2018;Ąđ?&#x2018;Ąđ?&#x2018;Ą(14.0 + đ?&#x2018;&#x2122;đ?&#x2018;&#x2122;), Minimize đ?&#x2018;&#x201C;đ?&#x2018;&#x201C;2 (x) = đ?&#x203A;żđ?&#x203A;ż(đ?&#x2018;Ľđ?&#x2018;Ľ) =

2.1952 đ?&#x2018;Ąđ?&#x2018;Ą 3 đ?&#x2018;?đ?&#x2018;?

of the material (30,000 psi). The đ?&#x2018;&#x201D;đ?&#x2018;&#x201D;3 makes certain that thickness of the beam is not smaller than the weld thickness from the standpoint. The đ?&#x2018;&#x201D;đ?&#x2018;&#x201D;4 keeps the allowable buckling load of the beam more than the applied load P for safe design. A violation of any of the above four constraints will make the design unacceptable. More on adjusting the constraints would be available in [22, 23]. Additionally, on the stress and buckling terms calculated in [24], it is worth mentioning that they are highly non-linear to design variables.

Subjected to đ?&#x2018;&#x201D;đ?&#x2018;&#x201D;1 (x) â&#x2030;Ą 13,600 â&#x2C6;&#x2019; đ?&#x153;?đ?&#x153;?(x) â&#x2030;Ľ 0,

đ?&#x2018;&#x201D;đ?&#x2018;&#x201D;2 (x) â&#x2030;Ą 30,000 â&#x2C6;&#x2019; đ?&#x153;&#x17D;đ?&#x153;&#x17D;(x) â&#x2030;Ľ 0, đ?&#x2018;&#x201D;đ?&#x2018;&#x201D;3 (x) â&#x2030;Ą đ?&#x2018;?đ?&#x2018;? â&#x2C6;&#x2019; â&#x201E;&#x17D; â&#x2030;Ľ 0,

đ?&#x2018;&#x201D;đ?&#x2018;&#x201D;4 (x) â&#x2030;Ą đ?&#x2018;&#x192;đ?&#x2018;&#x192;đ?&#x2018;?đ?&#x2018;? (x) â&#x2C6;&#x2019; 6,000 â&#x2030;Ľ 0

0.125â&#x2030;¤ â&#x201E;&#x17D;, bâ&#x2030;¤ 5.0,0.â&#x2030;¤ đ?&#x2018;&#x2122;đ?&#x2018;&#x2122;, tâ&#x2030;¤ 10.0

Creating the Model with Scilab The SciLab file contains a string definition, i.e. g_name, including a short, mnemonic name for the model as well as two 8-bit integers, i.e. g_dimension and g_range, defining the number of input and output variables of the model. Additionally, the file has two real-valued arrays; i.e. g_min and g_max, containing the minimum and maximum values allowable for each of the input and output variables. The following is a simple definition of a function that can be understood by utilized software package [59]. g_name = "ZDT1"; g_dimension = int8(2); g_range = int8(2); g_min = [0, 0, 0, 0]; g_max = [1, 1, 1, 1]; g_names = ["x1", "x2", "f1", "f2"]; function f = g_function(x) f1_x1 = x(1) g_x2 = 1 + 9 * x(2) h = 1 - sqrt(f1_x1 / g_x2) f = [ 1 - f1_x1, 1 - g_x2 * h ] endfunction; Among the four constraints, đ?&#x2018;&#x201D;đ?&#x2018;&#x201D;1 deals with the shear stress developed at the support location of the beam which is meant to be smaller than the allowable shear strength of the material (13,600 psi). The đ?&#x2018;&#x201D;đ?&#x2018;&#x201D;2 guarantees that normal stress developed at the support location of the beam is smaller than the allowable yield strength 20

FIG.5 DESCRIPTION OF THE WELDED BEAM DESIGN PROBLEM IN THE SOFTWARE ARCHITECTURE OF RSO MULTIOBJECTIVE

OPTIMIZATION; POINTING OUT THE OBJECTIVES AND CONSTRAINTS

In applying I-MODE framework to real-world design tasks of [22, 23] in which real decision makers will be involved there has a number of shortcomings which we needs further attention. Regarding the implementation of I-MODE software, a maximum of three objectives should be taken into account due to limitation of visual representation of the Paretooptimal solutions.

Setting up the RSO Software Here the RSO software architecture of LION solver [40, 33, 59] helps the designer to become aware of the different possibilities and focus on his preferred solutions, within the boundary of constraints.

Consequently, the constraints are transformed into a penalty function which sums the absolute values of the violations of the constraints plus a large constant.

Unless the two functions are scaled, the effect of deflection on the weighted sum will tend to be negligible, and most Pareto-optimal points will be in

International Journal of Information and Computer Science (IJICS) Volume 3 Issue 1, January 2014

www.iji-cs.org

the area corresponding to the lowest cost. Therefore each function is divided by the estimated maximum value of each function in the input range [24]. The Pareto-optimal solutions of the multiobjective optimization and MCDM corresponding to fabrication cost vs. end deflection of the beam are visually presented in the graph of Fig.6.

These observations can inspire a problem simplification, by fixing the height to its maximum value and by expressing the length as a function of depth, therefore eliminating two variables from consideration in the future explorations of this optimal design problem.

FIG.6 SET OF PARETO-OPTIMAL SOLUTIONS, FABRICATION COST VS. END DEFLECTION OF THE BEAM

FIG.8 MULTIDIMENTIONAL GRAPH FOR AN ADVANCED VISUALIZATION; THE FABRICATION COST VS. END DEFLECTION OF THE BEAM

By associating a multidimentional graph for an advanced visualization, available in Fig. 7, and a parallel chart, available in Fig. 8, to the results table, the MCDM problem very clearly comes to the consideration and the final decision is very confidently made. Further, it is observed that the welding length l and depth h are inversely proportional, the shorter the welding length is, the larger the depth has to be, and that height t tends to be close to its maximum allowable value [23].

FIG.7 PARELEL CHART INCLUDING ALL VARIABLES, CONSTRAINTS AND OPTIMIZATION OBJECTIVES

Conclusions In this paper, along with presenting two case studies, an interactive procedure is introduced which involves the DM in the optimization process helping to choose a single solution at the end. The proposed method is based on reactive search optimization algorithm and its novel software architecture packages, capable of handling the big data often associated with MCDM problems. According to case studies, the aspects of data mining, modeling, and visualization the data related to construction workers are considered.

Further, the utilization of the proposed software architectures for multiobjective optimization and decision-making, with a particular emphasis on supporting flexible visualization is discussed. The applicability of the software can be easily customized for different problems and usage contexts.

Considering the ability of LION solver and Grapheur, the interesting patterns are automatically extracted from our raw data-set via data mining tools.

Additionally, the advanced visual analytical interfaces are involved to support the DM interactively. With utilizing the features such as parallel filters and clustering tasks, in the construction project case study, the managers can solve MOO problems as it amends 21

www.iji-cs.org

International Journal of Information and Computer Science (IJICS) Volume 3 Issue 1, January 2014

previous approaches. Furthermore, the preliminary tests of the software environment in the concrete context of optimal designing, the welded beam design problem has shown the effectiveness of the approach in rapidly reaching a design preferred by the DM via advanced visualization tools and brain-computer novel interactions.

Author would like to thank Eesti Instituute for funding this research with the grant of Estophilus.

M. Esmaeili, A. Mosavi, “Variable reduction for multiobjective optimization using data mining techniques; application to aerospace structures,” in Proceedings of 2nd IEEE International Conference on Computer Engineering and Technology, Vol. 5. pp. 303-314, 2010.

A. Mosavi, “Application of multi-objective optimization packages in design of an evaporator coil,” World Academy of Science, Engineering and Technology, Vol. 61, No. 37, pp. 25-29, 2010.

[3]

A. Mosavi, “The multiobjective optimization package of IOSO; applications and future trends,” Conference of PhD Students in Computer Science, Szeged, Hungary, 2010.

[4]

A. Mosavi, “Computer design and simulation of built environment; application to forest,” Environmental and Computer Science, Vol. 1, pp. 81-85, 2009.

[5]

A. Mosavi, “Hydrodynamic design and optimization: application to design a general case for extra

[6]

February 2011.

[9]

A.Mosavi, “On engineering optimization the splined profiles,”

In

Proceedings

of

International

[10] A. Mosavi, “The large scale system of multiple criteria decision making; pre-processing,” Large Scale Complex Systems Theory and Applications, Vol. 9, No. 1, pp.

[11] A. Mosavi, “Applications of interactive methods of MOO in chemical engineering problems,” Global Journals of Engineering Research, Vol. 10, No. 3, Issue.

3, pp. 08-15, 2010.

[12] K. Miettinen, M.M. Makela "Interactive bundle-based method

for

nondifferentiable

multiobjective

optimization: NIMBUS", Optimization V34, 1995, 231– 246.

[13] A. Mosavi, “Decision-Making Models for Optimal Engineering

Design

and

their

Applications,”

Dissertation University of Debrecen.

[14] S. Chaudhuri, K. Deb, "An interactive evolutionary multi-objective optimization and decision making procedure", Applied Soft Computing, V10, 2010, 496– 511.

[15] J. Branke, "Consideration of Partial User Preferences in Evolutionary Multiobjective Optimization", Springer, Berlin, 2008, pp. 157–178.

[16] K.C.

Tan,

T.H.

Lee,

D.

Khoo,

E.F.

Khor,

"A

toolbox

for

multiobjective

Technology and Development, 2009. Vol. 2, pp. 139-143.

computer-aided multiobjective optimization", IEEE

A. Mosavi, “Parametric modeling of trees and using

Transactions on Systems, Man and Cybernetics—Part B:

integrated

Cybernetics 31 (4) (2001) 537–556.

CAD/CFD

and

optimization

tools:

evolutionay

algorithm

application to creating the optimal planting patterns for

[17] H. Takagi, "Interactive evolutionary computation:

new forests,” In Proceedings of 2nd International

fusion of the capabilities of ec optimization and human

Conference Wind Effects on Trees, Albert-Ludwigs-

evaluation", Proceedings of IEEE 89 (2001) 1275–1296.

R.

Battiti

and

optimization:

P.

Campigotto,

Learning

while

"Reactive

[18] K. Miettinen, "Nonlinear Multiobjective Optimization",

search

optimizing.

Kluwer, Boston, 1999.

an

[19] V.Chankong, Y.Y.Haimes, "Multiobjective Decision

experiment in interactive multiobjective optimization",

Making Theory and Methodology", North-Holland,

In S. Voss and M. Caserta, editors, Proceedings of MIC

22

From Data to Models to Reactive Search", Srl, Italy,

equipments on the submarine's hull,” Computer

University of Freiburg, Germany, pp. 213-222, 2009.

[7]

R. Battiti, M. Brunato, "Reactive Business Intelligence.

354-359, 2010.

REFERENCES

[2]

[8]

modeFRONTIER Users’ Meeting, Trieste, Italy, 2010.

ACKNOWLEDGMENT

[1]

2010.

New York, 1983.

2009, VIII Metaheuristic International Conference,

[20] A.P. Wierzbicki, "The use of reference objectives in

Lecture Notes in Computer Science. Springer Verlag,

multiobjective optimization", in: G. Fandel, T. Gal (Eds.),

International Journal of Information and Computer Science (IJICS) Volume 3 Issue 1, January 2014

www.iji-cs.org

and

[30] A. Mosavi, et al., “Reconsidering the multiple criteria

Applications, Springer-Verlag, Berlin, 1980, pp. 468–486.

decision making problems of construction workers with

[21] R. Kamalian, H. Takagi, A. Agogino, "Optimized design

the aid of grapheur,” International ANSYS and

Multiple

Criteria

Decision

Making

Theory

of mems by evolutionary multi-objective optimization with

interactive

Proceedings

of

evolutionary the

computation",

Genetic

and

in:

Evolutionary

Computing Conference (GECCO), 2004, pp. 1030–1041.

[22] K. Deb, H. Gupta, "Searching for robust Pareto-optimal solutions

in

multi-objective

optimization",

in:

Proceedings of the Third Evolutionary Multi-Criteria

EnginSoft Conference, Verona, Italy, 2011.

[31] C.V.

Jones,

"Feature

Article–Visualization

and

Optimization". INFORMS Journal on Computing, 6, pp.

221-229 (1994).

[32] P. Piero, R. Subbu, J. Lizzi, "MCDM: A framework for research

and

applications".

IEEE

Computational

Intelligence Magazine. 4, pp. 48–61(2009).

Optimization (EMO-05) Conference (Also Lecture

[33] R. Battiti, M. Brunato, F. Mascia, "Reactive Search and

Notes on Computer Science, vol. 3410), 2005, pp. 150–

Intelligent Optimization". Operations research/Computer

164.

Science Interfaces. Springer Verlag, (2008).

[23] K. Deb, S. Chaudhuri, "I-MODE: an interactive multi-

[34] A.

Mosavi,

“Multiple

criteria

decision-making

objective optimization and decision-making using

preprocessing using data mining tools,” International

evolutionary methods", in: Proceedings of Fourth

Journal of Computer Science Issues, Vol. 7, Issue. 2, No.

International

1, pp. 26-34, 2010.

Conference

on

Evolutionary

Multi-

Criteria Optimization (EMO 2007), 2007, pp. 788–802.

[24] A. Mosavi, Optimal Engineering Design. Tech. Rep.

[35] A. Adejuwon, A. Mosavi, “Domain driven data mining; application to business,” International Journal of Computer Science Issues, Vol. 7. No. 4, pp. 41-45, 2010.

2013. University of Debrecen, Hungary, 2013.

[25] A. Mosavi, A. S. Milani, M. Hoffmann and M. Komeili,

[36] A.

Mosavi,

A.

Vaezipour.

"Reactive

Application

to

Search

“Multiple criteria decision making integrated with

Optimization;

Multiobjective

mechanical modeling of draping for material selection

Optimization Problems." Applied Mathematics 3, no. 30

of textile composites,” In Proceedings of 15th European

(2012): 1572-1582.

Conference on Composite Materials, Italy, Venice, 2012.

[37] A. Mosavi, “Multiobjective optimization of Spline

[26] A. Mosavi, M. Hoffmann, A. S. Milani, "Adapting the

curves using modeFRONTIER,” In Proceedings of

Reactive

Search

Optimization

and

Visualization

Algorithms for Multiobjective Optimization Problems; Application to Geometry", Conference of PhD Students

International modeFRONTIER Users’ Meeting, Trieste, Italy, 2010.

[38] G. Anzellotti, R. Battiti, I. Lazzizzera, P. Lee, A. Sartori,

in Computer Science, Szeged, Hungary, June, 2012

G. Soncini, G. Tecchiolli, A. Zorat, "Totem: a highly

[27] A. Mosavi, M. Hoffmann and A. Milani, “Optimal

parallel chip for triggering applications with inductive

design of the NURBS curves and surfaces utilizing

learning based on the reactive tabu search", In:

multiobjective

optimization

and

decision

making

algorithms of RSO,” Conference of PhD Students in Mathematics, Szeged, Hungary, 2012.

[28] E. Foldi, et al. “Reconsidering the multiple criteria decision making problems of construction projects;

AIHENP95. Pisa, IT 1995.

[39] R. Battiti, M. Brunato, "Reactive search optimization: learning

while

optimizing",

In:

Handbook

of

Metaheuristics, 2nd edn. Springer, Berlin, 2009.

[40] R. Battiti, M. Brunato, F. Mascia, "Reactive Search and

using advanced visualization and data mining tools,”

Intelligent

Conference of PhD Students in Computer Science,

research/Computer Science Interfaces, vol. 45. Springer

Szeged, Hungary, 2012.

Verlag, 2008.

Optimization",

Operations

[29] A. Mosavi, et al., “Reconsidering the multiple criteria

[41] R. Lenne, C. Solnon, T. Stutzle, E. Tannier, M. Birattari,

decision making problems of construction workers;

"Reactive Stochastic Local Search Algorithms for the

using grapheur,” ENGINSOFT Newsletter, Year. 8, No.

Genomic Median Problem" LECTURE NOTES IN

4, 2011.

COMPUTER SCIENCE 4972, 266 (2008).

23

www.iji-cs.org

International Journal of Information and Computer Science (IJICS) Volume 3 Issue 1, January 2014

[42] T. Genji, T. Oomori, K. Miyazato, N. Hayashi, Y.

self-adaptive neighborhoodordering", In: C. Cotta, A.J.

Fukuyama, K. Co, "Service Restoration in Distribution

Fernandez, J.E. Gallardo (eds.) Proceedings of the 7th

Systems Aiming Higher Utilization Rate of Feeders", In:

EU/MEeting on Adaptive, Self-Adaptive, and Multi-

Proc.

Level Metaheuristics, malaga, Spain, 2006.

of

the

Fifth

Metaheuristics

International

Conference (MIC2003), 2003.

[54] J. Ryan, T. Bailey, J. Moore, W. Carlton, "Reactive tabu

[43] R. Russell, W. Chiang, D. Zepeda, "Integrating multi-

search in unmanned aerial reconnaissance simulations",

product production and distribution in newspaper

Proceedings of the 30th conference onWinter simulation

logistics", Computers and Operations Research 35(5),

pp. 873–880, 1998.

1576–1588, 2008.

[55] R. Kincaid, K. Laba, "Reactive Tabu Search and Sensor

[44] W. Nanry, J. Wesley Barnes, "Solving the pickup and

Selection

inActive

Structural

Acoustic

Control

delivery problem with time windows using reactive

Problems", Journal of Heuristics 4(3), 199–220, 1998.

tabu search", Transportation Research Part B 34(2), 107–

[56] P. Hansen, N. Mladenovic, "Variable neighborhood

121, 2000.

search", In: E. Burke, G. Kendall (eds.) Search

[45] A. Login, S. Areas, "Reactive tabu adaptive memory

methodologies: introductory tutorials in optimization

programming search for the vehicle routing problem

and decision support techniques, pp. 211–238. Springer,

with backhauls", Journal of the Operational Research

2005.

Society 58, 1630–1641, 2007.

[57] K.

Hamza,

H.

Mahmoud,

K.

Saitou,

"Design

[46] J. Chambers, J. Barnes, "New tabu search results for the

optimization of N-shaped roof trusses using reactive

job shop scheduling problem", The University of Texas,

taboo search", Applied Soft Computing Journal 3(3),

Austin,

221–235, 2003.

Graduate

TX,

Technical

Program

in

Report

Series

Operations

ORP96-06,

Research

and

Industrial Engineering , 1996.

[58] J. Błachut, "Tabu search optimization of externally pressurized

[47] H. Delmaire, J. Dıaz, E. Fernandez, M. Ortega, "Reactive

barrels

and

domes",

Engineering

Optimization 39(8), 899–918, 2007.

GRASP and Tabu Search based heuristics for the single

[59] Brunato, Mauro, and Roberto Battiti. "Grapheur: a

source capacitated plant location problem", INFOR 37,

software architecture for reactive and interactive

194–225, 1999.

optimization." Learning and Intelligent Optimization.

[48] A. Fink, S Voß, "Solving the continuous flow-shop scheduling problem by metaheuristics". European Potocnik,

I.

Grabec,

"Adaptive

[60] B. Roberto, A. Passerini (2010). "Brain-Computer Evolutionary Multi-Objective Optimization (BC-EMO):

Journal of Operational Research 151(2), 400–414, 2003.

[49] P.

Springer Berlin Heidelberg, 2010. 232-246.

self-tuning

a genetic algorithm adapting to the decision maker."

in

(PDF). IEEE Transactions on Evolutionary Computation

[50] T. Winter, U. Zimmermann, "Real-time dispatch of

[61] Hewage K.N., Gannoruwa A., Ruwanpura J.Y. (2011),

trams in storage yards", Annals of Operations Research

Current Status of Factors Leading to Team Performance

(96), 287–315 (2000).

of On-Site Construction Professionals in Alberta

neurocontrol",

Mathematics

and

Computers

14 (15): 671–687.

Simulation 51(3-4), 201–207 (2000)

[51] M. Magdon-Ismail, M. Goldberg, W. Wallace, D. Siebecker, "Locating hidden groups in communication networks using hidden markov models" LECTURE NOTES IN COMPUTER SCIENCE pp. 126–137 (2003) [52] M. Hifi, M. Michrafy, A. Sbihi, "A Reactive Local Search-Based Algorithm for the Multiple-Choice MultiDimensional

Knapsack

Problem",

Computational

Optimization and Applications 33(2), 271–285, 2006.

[53] B. Hu, G.R. Raidl, "Variable neighborhood descent with

24

Building Construction Projects, Canadian Journal of Civil Engineering.

[62] Hewage K.N., Ruwanpura J.Y., Jergeas G.F. (2009). IT Usage in Alberta’s Building Construction Projects: Current Status and Challenges, Journal of Automation in Construction.

[63] J. March, Bounded rationality, ambiguity, and the engineering of choice, The Bell Journal of Economics, pp. 587-608, 1978.

International Journal of Information and Computer Science (IJICS) Volume 3 Issue 1, January 2014

[64] A. Vaezipour and A. Mosavi, “Managing decision making within enterprise,” Poster in Proceedings of

www.iji-cs.org

material selection,” ENGINSOFT newsletter, simulation based engineering & Sciences, No. 4, Winter 2012.

[75] A. Mosavi, “A MCDM software tool for automating the

International CAE Conference, Verona, Italy, 2012.

[65] A. Mosavi and A. Vaezipour, “Enterprise decision

optimal design environments, with an application in

management with the aid of advanced business

shape optimization,” In Proceedings of International

intelligence” Conference

In on

proceedings Computer

of

International

Conference on Optimization and Analysis of Structures,

Science,

Engineering,

Tartu, Estonia, 2013.

Technology and Application (ICCSETA), Budapest,

[76] A. Mosavi and A. Adeyemi, “On domain driven data mining and business intelligence,” 8th Joint Conference

Hungary, 2012.

[66] A. Mosavi, M. Hoffmann and A. S. Milani, “Adapting the reactive search optimization and visualization

on Mathematics and Computer Science, Komarno, Slovakia, 2010.

algorithms for multiobjective optimization problems;

[77] A. Mosavi, “Data mining for business applications,” 3rd

application to geometry,” Conference of PhD Students

International Symposium on Business Information

in Computer Science, Szeged, Hungary, 2012.

Systems, Pecs, Hungary, 2010.

[67] A. Mosavi, “Brain-computer optimization for solving

[78] A. Mosavi, “Data mining for business applications and

complicated geometrical decision-making problems,” In

business decision-making: challenges and future trends,”

Proceedings of PEME VI. Ph.D. Conference, Budapest,

3rd international Symposium on Business Information

Hungary, 2013.

Systems, Pecs, Hungary, 2010.

[68] A. Mosavi, “A MCDM software tool for the automated

[79] A. Mosavi, “Hydrodynamic design optimization,” In

design environments,” In Proceedings of 26th Europian

Proceedings of 15th International Conference on

Conference on Operational Research, Rome 2013.

Building Services, Mechanical and Building Industry

[69] A. Mosavi, “Visualization-based software architecture for

decision

making,”

International

Journal

of

Computer Science and Application, 2013.

Days, Debrecen, Hungary, 2009.

[80] A. Mosavi, M. Hoffmann and N. Peter, “Automatic multi-objective

[70] A. Vaezipour and A. Mosavi, “Enterprise decision

surface

modeFRONTIER's

design

optimisation

CAD/CAE

integrated

using system:

management with the aid of advanced business

Application to military submarine sail,” EnginSoft

intelligence and interactive visualization tools,” in

International

Proceedings of International CAE Conference, Verona,

Conference, Bergamo, Italy, 2009.

Italy, 2012.

Conference

and

ANSYS

Italian

[81] A. Mosavi, “Application of multi-objective optimization

[71] A. Mosavi, “Decision-making software architecture; the

packages in coupling ANSYS with CAD packages and

visualization and data mining assisted approach,”

EXCEL,” In Procedings of ANSYS Conference & 27.

International Journal of Information and Computer

CADFEM users’ meeting, Congress Center Leipzig,

Science, 2013.

Germany, 2009.

[72] A. Mosavi, “On developing a decision-making tool for

[82] A.

Mosavi,

“Multi-objective

shape

optimization;

general applications to computer vision,” International

application to design a thermal-fluid structure,” In

Journal of Computer Applications, Special Issue on

Proceedings of Third International Conference on

Recent Trends in Pattern Recognition and Image

Multidisciplinary

Analysis RTPRIA(1): pp. 10-17, 2013.

Applications, Paris, France, pp. 77-82, 2010.

Design

Optimization

and

[73] A. Mosavi, et al., “Multiple criteria decision making

[83] A. Mosavi, “Data mining for decision making in

integrated with mechanical modeling of draping for

engineering optimal design,” Journal of Artificial

material selection of textile composites,” Poster In

Intelligence & Data Mining, Vol. 1, 2013. In Press.

Proceedings

of

15th

European

Conference

on

Composite Materials, Italy, Venice, 2012.

[74] A. Mosavi, M. Hoffmann, A. Vaezipour, “Grapheur for

[84] A.

Mosavi,

“Application

of

data

mining

in

multiobjective optimization problems,” International Journal for Simulation and Multidisciplinary Design

25

www.iji-cs.org

International Journal of Information and Computer Science (IJICS) Volume 3 Issue 1, January 2014

Optimization.

412-422, 2010.

[85] A. Mosavi and M. Hoffmann, “Design of curves and

[89] A. Mosavi, “Multiobjective optimization using indirect

surfaces by multiobjective optimization; utilizing IOSO

optimization on the basis of self-Organization,” In

and modeFRONTIER packages,” Poster in Proceedings

Proceedings

of

Computational Intelligence, Bucharest, Romania, 2010.

Enginsoft

International

Conference

on

CAE

Technologies for Industries, Bergamo, Italy, 2010.

of

International

Conference

on

[90] A. Vaezipour, et al., “Machine learning integrated

[86] A. Mosavi, “Interactive methods of MOO; application

optimization for decision making,” In Proceedings of

to chemical engineering problems,” In Proceedings of

26th Europian Conference on Operational Research,

Third International Conference on Multidisciplinary

Rome 2013.

Design Optimization and Applications, Paris, France, pp. 67-76, 2010.

Supports

Tough

Decisions

within

Construction Projects, International CAE Conference, 2011.

[88] A. Mosavi, “Multiobjective optimization package of IOSO,” In Proceedings of 24th Mini EURO Conference on Continuous Optimization and Information-Based Technologies in the Financial Sector, Izmir, Turkey, pp.

26

Mosavi,

“Decision-Making

Models

for

Optimal

Engineering Design and their Applications,” Dissertation

[87] A. Mosavi, KN Hewage, AS Milani, M Yeheyis, Grapheur

[91] A.

University

of

Debrecen,

2012.

hdl.handle.net/2437/171847.

[92] A.

Mosavi,

“A

multicriteria

decision

making

environment for engineering design and production decision-making,” International Journal of Computer Applications, Vol. 69, No. 1, pp. 26-38, 2013.