طراحی سلسله مراتبی استدلال مبتنی بر مورد در کاربرد کارت امتیازی متوازن

A balanced scorecard (BSC) is a management decision tool intended to be the corporate performance measurement. It also can play an important role in transforming an organization’s mission and strategy into a balanced set of integrated performance measures. Assigning suitable weight to each level of balanced scorecard is crucial to conduct performance evaluation effectively. In this research a case-based reasoning (CBR) system has been developed to assist in assigning the suitable weights. Based on the balanced scorecard design, this study proposed a three-level feature weights design to enhance CBR’s inference performance. For effective case retrieval, a genetic algorithm (GA) mechanism is employed to facilitate weighting all of levels in balanced scorecard and to determine the most appropriate three-level feature weights. The proposed approach is compared with the equal weights approach and the analytical hierarchy process (AHP) approach. The results indicate that the GA-CBR approach is able to produce more effective performance measurement.

The BSC (Kaplan & Norton, 1992) is a performance measurement framework to allow managers to look at their business performance from four performance perspectives – financial, customer, internal business and innovation and learning. The weight of each feature in balanced scorecard is an impact factor to evaluate performance. The AHP method (Saaty, 1990) is used to generate the weights (Stewart & Mohamed, 2001). The AHP method is often used as an effective tool in structuring and modeling multi-criteria problems because it attempts to quantify human judgment and opinion that other approaches might ignore. However, by using pair-wise comparison the calculation of preference between criteria is mainly based on some quantitative business data and the subjective judgment from senior management level. No matter how professional they are, the results based on the judgment of those decision-makers somewhat are subject to imprecision. A proposed CBR approach is suggested to handle subjective judgment problems. CBR is a machine reasoning that adapts previous similar cases to infer further similarity. It allows a computer program to propose solutions in domains that are not completely understood (Kolodner, 1992). To develop a CBR system, a set of useful case features must first be determined to differentiate one case from the others. Furthermore, weights representing the importance of features have to be assigned in the case-matching process. In order to apply CBR to the balanced scorecard, this study adopts the CBR system with a three-level weight design. The weights are usually determined by subjective judgments or the trial and error approach. Instead of subjective judgments or the inefficient way of trial and error, the GA is adopted to determine the weights (Chiu, 2002). To evaluate business performance, this study presents a CBR system with the GA which is used to determine the optimal three-level weights. Without any human judgments with questionnaires, such as AHP method, the weight can be produced automatically by computer. Then, the weight produced can be used in balanced scorecard to evaluate performance.

This paper presented the GA-CBR approach to generate weights that can be used in balanced scorecard to evaluate performance. CBR is a general problem-solving method, particularly useful in retrieving similar cases from historical operation records. When faced with a new problem, it searches from similar cases from the case base and reuses the solutions of the similar cases in the current problem. This research investigates the impacts of both feature weighting and similarity functions for improving the case retrieval effectiveness. Instead of subjective judgments or the inefficient way of trial and error, the GA is adopted to enhance the case-matching process for determining a three-level weights design of CBR for performance evaluation. Defining appropriate weighting values for each feature is a crucial issue for effective case retrieval. This paper proposed the GA-based approach to determine the fittest weighting values for improving the case evaluation accuracy. Compared with AHP and equal weights approach, our proposed GA-CBR approach without human judgments really improves the weight accuracy. It can help managers in doing performance evaluation more efficiently and correctly. Otherwise, once the model is established, the weights can be generated easily by computer automatically without any questionnaires and decision-makers’ judgments. The managers can review, evaluate, and justify their decisions through intranet, which can streamline the performance evaluation process, then helps e-business growth. The proposed GA-CBR approach that is based on learning the historical cases is objective as it minimizes the effect of subjective factors that has to be carefully dealt in performance evaluation process. Because the similarity functions may influence the case association process, future research may work on different combinations of similarity functions between case features based on balanced scorecard to examine their retrieval effectiveness.