An Approach towards Enterprise Architecture Analysis using AHP and Fuzzy AHP

Abstract

The related work to this paper consists of 3 groups: 1. Software quality attributes measurement methods based on MCDM methods [10], [11], [12], [13], [14] we have used the idea of some of these methods in our analysis approach. 2. The analysis methods in the EA community including [2] [15], [16], [17], [18], [19], [20], [21], [22], [23], [24]. The main contributions of our approach which make it different from the approaches of group 2 are as below: a) In our approach, the criteria and sub-criteria of a quality attribute are given different weights according to the EA layers each of them belong to and also the importance of each EA layer in the enterprise. Above mentioned approaches, use causal effect and probability theory, and model causal probabilities between quality attributes and criteria. b) Through our approach, we use the knowledge and experience of two groups of experts in our assessment; EA experts and domain experts. This ensures a broader decision base according to different points of view and allows identification of differences in experiences. c) All above methods use formal languages such as Influence Diagrams or their extended version to support the analysis of EA, but we have used Analytical Hierarchy Process (AHP) as a multi criteria decision making method, which is the first experience of using this method in the field of EA analysis. 3. There are some similar EA analyses approaches which are based on the same idea of this paper; References [25][26] and the most completely presented approach [27] have not used fuzzy in the EA analysis approach. Also [28] contains the idea of using fuzzy AHP in the EA analysis approach but the approach is less mature with less detail. III. THE PROPOSED METHOD The aim of this paper is to provide a method to facilitate deciding between different scenarios according to their level of quality attribute achievement. This method is usable after gathering complete information about the current EA of the enterprise. In this approach, each quality attribute is composed of several criteria and sub-criteria which are specified in the quality attribute general scenario [3]. This includes subjective judgment with or without consensus building and methods such as providing a total sum of points to be divided between the items you would like to prioritize. This opportunity arises from the fact that AHP is based on all pair-wise comparisons. To better understand the proposed approach, we first precisely describe AHP and Fuzzy AHP methods used in this paper, and then explain the proposed approach in a step by step manner (Fig. 1). In this figure the yellow boxes represent the steps that directly use AHP and the green box represent the step that uses Fuzzy AHP. A. Analytical Hierarchy Process (AHP) and Fuzzy AHP AHP consists of a set of steps, where all combinations of elements are evaluated pair-wise, and according to a certain scale (Fig. 2). The question to answer for each pair-wise comparison is which of the two elements, i or j is more important, and how much more important is it? This is rated by interpreting the values as presented in Table I. For more detail about AHP approach, please refer to [27] [28] [4] [5]. In this study, triangular fuzzy numbers, 1 % to 9 % , are used to represent subjective pair-wise comparisons of selection process (equal to extremely preferred) in order to capture the vagueness and imprecision of human qualitative assessments.