WP4: Multi-Criteria Analysis for Software Process Assessment/Planning

Duration M14 – M23

Description of the approach: This WP will investigate the combined application of multi-criteria decision analysis techniques to support the planning phase of a process assessment/improvement project. The effectiveness of any possible approach will be examined from a number of perspectives, such as the organization needs, the cost/risk of application/deployment, the value/benefit of each possible improvement issue, the time to return on investment, the need to reach a high-degree of agreement among all stakeholders involved upon the final improvement plan etc.

In particular, one of the main difficulties when performing prioritization of requirements in a software process improvement project is the fact that requirements are derived from different stakeholders with different priorities and needs (e.g., assessors/evaluators, customers, end-users, project managers, requirements engineers, developers, testers etc.) [37, 38]. A key factor for a successful implementation is to effectively negotiate the requirements among these stakeholders who have different priorities and needs. There is need to manage risks earlier and identify the most important requirements for satisfying as much as possible all stakeholders.

The variety of alternative improvement solutions, especially for a RE process improvement, along with the variation among each organization needs, make the determination of an improvement plan (that is a sequence of improvement steps/practices) a complicate multi-criteria decision problem that needs to consider viewpoints of all stakeholders. The problem is often approached in practice based only on personal preferences of the involved project managers or process evaluators. Such an approach may lead to improvement plans of high-cost and final suggestions which do not reflect the actual organization requirements.

To meet these problems, in this WP we will examine powerful techniques from the multi-criteria decision making (MCDM) domain. Although, MCDM methods like AHP and QFD have been applied for the prioritisation of SW improvement issues (e.g., [1, 37, 38]), we believe that the problem requires more powerful techniques to deal with the vagueness or even the unawareness of stakeholders to express the importance of their requirements or to rate their judgements. Treating with these ambiguities and consideration of incomplete available information expose the need to adopt a fuzzy-based approach [26, 27, 28]. Fuzzy-based methods provide the intuitive advantage to utilize, instead of crisp values, linguistic terms to evaluate performance of the alternatives and criteria weights.

A fuzzy-based approach can be even more beneficial when it is combined with other decision making techniques. For example, fuzzy AHP [26] is proposed to handle the inherent imprecision in the pairwise comparison process, while fuzzy TOPSIS (Technique for Order Preference by Similarity to Ideal Solution) [27] can be used to jointly consider both positive (benefit) and negative (cost/effort) criteria. Therefore, in this WP we have plans to examine the application of these techniques to support decision making within the context of software process improvement.

An innovative aspect in WP4 is to investigate the adoption of intuitionistic fuzzy sets, an extension of fuzzy sets proposed by Atanassov [28] that has been used mainly in decision making problems of supplier selection [29, 30, 31]. An intuitionistic fuzzy set includes the membership and the non-membership function of an element to a set as well as a third function that is called the hesitation degree. This third function is useful to express lack of knowledge and hesitancy concerning both membership and non-membership of an element to a set. Expression of hesitation is particularly helpful for representing indeterminacy of assessors/evaluators, end-users and members of the SW organization when they select which steps/practices of a software process are required to be adopted and/or improved. Therefore, our aim is to investigate the combined use of intuitionistic fuzzy sets with TOPSIS. The advantage of this combination, in case of software process assessment, is that provides means to distinguish between positive (e.g., business benefit, profit, product quality, ease of implementation) and negative (e.g., cost, effort, risk) criteria for the selection and prioritisation of software process requirements.

Deliverables:

D4.1: Multi-Criteria Analysis for Software Process Assessment/Planning (M14-M23)

WP4 Leader: Pandelis Ipsilandis

WP4 Members: V.C. Gerogiannis, D. Tselios, E. Rapti

Top