Framework for Optimal Selection Using Meta‐Heuristic Approach and AHP Algorithm

Abstract

Many real-life decisions are focused on selecting the most preferable combination of available options, by satisfying different kinds of preferences and internal or external constraints and requirements. Focusing on the well-known analytical hierarchical process (AHP) method and its extension CS-AHP for capturing different kinds of preferences over two-layered structure (including conditionally defined preferences and preferences about dominant importance), we propose a two-layered framework for identifying stakeholders' decision criteria requirements and employ meta-heuristic algorithms (i.e., genetic algorithms) to optimally make a selection over available options. The proposed formal two-layered framework, called OptSelectionAHP, provides the means for optimal selection based on specified different kinds of preferences. The framework has simultaneously proven optimality applied in software engineering domain, for optimal configuration of business process families where stakeholders' preferences are defined over quality characteristics of available services (i.e., QoS attributes). Furthermore, this domain of application is characterized with uncertainty and variability in selection space, which is proven and does not significantly violate the optimality of the proposed framework.

Multi-Criteria Decision Making for Optimal Configuration of Business Process Model Families

“…[20]. Key challenge in modelling and processing user preferences is posed by the need to express human opinion in a way that can be easily processed by computers [28]. Existing approaches could be categorized based on both, different kinds of preferences, input scales and quality measurements they are generating, as summarized in Table 1.…”

“…To ascertain the degree to which combination of services is appropriate for a stakeholder, we will elicit different kinds of stakeholders' preferences over QoS values Q = {q 1 , …, q k } by extending the approach developed in [28], which includes the following steps (see Fig. 4).…”

“…3.1.2. Specification of Stakeholders' Preferen Form of Two-Layered Structure over BPMF To ascertain the degree to which comb services is appropriate for a stakeholder, we different kinds of stakeholders' preferences values Q = �q � , … , q � � by extending the developed in [28], which includes the follow (see Fig. 4):…”

“…With reference to previous work [15,21,28] as well as to developed fuzzified measurements over CS-AHP two-layered structure, the fitness function will incorporate two types of penalty factors, as follows:…”

“…-Fitness evaluation-A fitness function is used for assessing each member of population as well as making performance analyses over the population. With reference to previous work [15,21,28] as well as to developed fuzzified measurements over CS-AHP twolayered structure, the fitness function will incorporate two types of penalty factors, as follows: -Penalty factor 1: weighted distance from constraint satisfaction measured as: ,..., ,..., ,…”