**2. Running example**

preferences is a great challenge, as it is difficult to express human opinion in a way that can be easily processed by computers [1]. Researchers in many different fields (such as, economics, riskmanagement,decisiontheory,socialchoicetheory,operationalresearch,intelligentsystems, databases, etc.) studied the representation of preferences, its processing and practical use [2].

194 Applications and Theory of Analytic Hierarchy Process - Decision Making for Strategic Decisions

Additionally, there is one more demand on allowing automated support for selection of the most preferable combination of available options, with respect to specified preferences and constraints if exist. In order to develop a framework for both, representing and reasoning about different kinds of requirements, the following two issues should be carefully analyzed: (i) comprehensive model for presentation of different kinds of preferences, on which bases develop (ii) the approach for automated selection by optimizing the fitting degree of satisfying

Traditional elicitation methods are typically developed based on pair‐wise comparisons, priority groups, networks for decision‐making and cumulative ratings [3, 4]. They usually collect independent preferences, under the mutual preference independence (MPI) hypothesis [5], which means that a user's preference for an option is independent of the other options [1]. However, the MPI hypothesis is not always true in practice [6] since people often express conditional preferences as to be more natural to the human way of thinking [1]. This is why well‐accepted and widely used analytical hierarchical process (AHP) algorithm proposed by Saaty [7] has been recently extended in order to handle conditionally defined preferences over

On the other hand, there is a wide range of different optimization and search techniques that have been used for solving the optimization problems [8]. Classical techniques [such as linear programming (LP)] are often distinguished as straightforward deterministic algorithms which are distinct from meta‐heuristic search, such as, hill climbing [9], simulated annealing [10] and genetic algorithms (GAs) [11]. However, these deterministic optimization algorithms are often inapplicable in many real‐world problems, because the problems have objectives that cannot

In this chapter, we demonstrate how the selection processes can be automated in a more scalable manner by using AHP (and its extension for handling conditionally defined prefer‐ ences, CS‐AHP) and Genetic Algorithms (GA) for presentation and analyses of different kinds of preferences and solving the problem of the optimal selection over the set of available options. Our framework, called *OptSelectionAHP*, provides the following major benefits to the process

**1.** It proposes an adoption of CS‐AHP method that enables its use for both, capturing and handling different kinds of preferences, as well as definition of optimal selection goals

**2.** It proposes the use of genetic algorithms adapted to quality measurements defined on the bases of CS‐AHP outputs over created two‐layered selection criteria structure;

**3.** It is able to effectively solve optimal selection problems and its optimality is not affected

by the presence of uncertainty and variability in the selection space.

specified preferences.

two‐layered hierarchical structure, namely CS‐AHP [2].

be characterized by a set of linear equations [12].

of prioritization, decision making and optimal selection:

over two‐layered selection criteria structure;

In order to exemplify the whole approach, we analyze simplified example of persons planning how to spend annual budget for vacations and other expenditures that are not ordinarily calculated (e.g., concerts of famous rock groups, theaters, sport events, etc.).

Person A is aware of total budget expenditure which cannot be exceeded (e.g., 7000) and interested for at least one of three demands (summer/winter holidays and rock concert). Several options are available for each demand, such as summer holidays can be spent in own country, or in the famous holiday resorts, or at a luxury destination. On the other side, three different options are also available for winter holidays: near winter resort, famous international winter resort and luxury destination. He/She is interested to attend two different concerts: one in own city, and another in the neighbor country (which requires additional traveling costs).



**Figure 1.** Illustrative example: (a) available options characterized in accordance with selection criteria, (b) defined con‐ straints and (c) aggregated range values for selection criteria.

Even that money is a key limitation factor, person A would like to be satisfied with fulfillment of other personal attitudes and preferences, such as high comfort, low traveling costs, but in the case of medium comfort, he/she accepts to spend money on higher travel costs.

On the other side, person B could spend between 3000 and 4000 for the same demands (summer and winter holidays and rock concerts), and he/she is not highly interested for comfort, and he/she would like to see as many famous destinations as possible. Also, person B is interested to both have at least one holiday and attend the concert. Available options and values of key decision criteria are illustrated in **Figure 1**.
