Software Component Selection Based on Quality Criteria Using the Analytic Network Process

and Applied Analysis 3 Panagiotou and Mentzas used KnowBench that supports the knowledge management process for the designing and implementation of sotware [34]. Khan and Mahmood have proposed a component selection process that uses a graph model which in turn is signed for interdependencies of CBS and group related goals into clusters [35]. However, no previous work has been done on the selection of sotware components based upon attributes of quality criteria when there are dependencies among the attributes. Hence, to overcome this limitation of dependencies among attributes, the proposed ANP method is suggested that is well-organized and incorporates the quality attributes of ISO/IEC 25010:2011 [36]. 3. Analytic Network Process for Software Component Selection he motivation of the ANP method toward selection of sotware components is to the dependencies existing among attributes of the network of elements. he ANP method is very eicient in dealing with such phenomena. ANP has various applications such as [37–39]. he details on ANP can be found in Saaty (1996) [10]; however, the main steps are summarized as follows. (1) he problem is divided into a network of subproblems. (2) A qualitative scale of importance which is presented by Saaty is as follows: (equal importance allocate similarity to objectives), moderately important (somewhat good turn action above a new), reasonable plus, strong weight (powerfully good turn one action above another), strong plus, verywell-built conirmed important (very powerfully favor above another), very, very strong, and excessive importance are given. he scale is converted into a quantitative scale of range between 1 and 9. (3) Pairwise comparison is done in step (3). he criteria in the “i” row are compared with the criteria in the “j” column in the form of (� �). If the criteria of “I” row are superior to “j” column, then it is written as (� �), and (� �) is the reciprocal of (� �). (4) he relative importance is calculated by inding the principal eigenvalue and the related eigenvector of the comparisonmatrix.he elements are normalized and are termed as weights of the criteria or subcriteria. (5) Ater pairwise comparisons the consistency of the matrix is measured. Priority vector “�” is calculated as follows: �� = �max�, (1) where �max is the major eigenvalue of matrix “�” and “�” is its eigenvector. he value of “�” is obtained by summing the column values of the original matrix multiplied by the normalized eigenvector.heprincipal eigenvector is obtained by the sum of all “�.” Table 1: Random consistency index. � 1 2 3 4 5 6 7 8 9 RI 0 0 0.6 0.9 1.1 1.2 1.3 1.4 1.5