Framework for Aggregating Interactive Multi-Criteria for Web-Based Applications Using Fuzzy Measure and 2-Additive Choquet Integral

Web-based Application (WBA) quality is hard to evaluate because it consists of multiple criteria to be measured. Researchers have developed software quality models that were intended to be comprehensive and applicable to all software development. However, most of the works that focus on the quality model do not reference the aggregation methods used to calculate the values of the different element of the model. In fact, the aggregation methods are not really discussed in the literatures related to software quality. Although the need for a sound methodology is widely recognised, previous researchers generally avoid the use of multi-criteria decision method and mostly dealt with the simplest form of aggregation, which is categorized under additive approach. This approach presents some drawbacks, some do not posses all desirable properties and some seem to be restrictive. As a result they tend to construct independent criteria, or criteria that are supposed to be so, which caused some bias effect in the evaluation. The purpose of this research is to establish a multi-criteria aggregation framework to evaluate WBA that would allow evaluator to incorporate interaction behaviour in the aggregation process. Firstly, the relationships between quality criteria were investigated using survey and correlation studies. The studies have confirmed that there exist positive, negative or neutral relationships between software quality criteria. Based on the relationships identified, the interactions of quality criteria are modelled using fuzzy measure approach. Secondly, the produced fuzzy measures are integrated into the 2-additive Choquet Integral to aggregate the multi quality criteria values. Besides that, the relative importance of each quality factor and the Web preference also play an important role in determining the overall evaluation. Case studies of three different domains of WBA were conducted to validate the proposed aggregation procedure. The overall results of 2-additive Choquet integral are compared against the additive model approach. Consequently, results of the 2-additive Choquet integral have shown that the ranking of overall evaluation results reflect the earlier preference stated by the evaluator compared to the additive model approaches. The fuzzy measure modelling has allowed the incorporation of interaction exist between the quality criteria to be considered in the aggregation process. The results are also consistent with the additive model approach when no interaction exists