WP2: Requirements Analysis for SPIRE Projects

Duration: M1-M08

Description of the approach: During WP2 a detailed requirements analysis will be performed for defining and analyzing the success factors of a process assessment/improvement project. Emphasis will be given on identifying process improvement requirements in the context of requirements engineering activities which take place into representative and very dynamic Greek SW SMEs.

Measurement of the success and quality of a RE process, in general, is a difficult task for the following reasons [47, 48, 49]. First, RE is an inherently complex and non-standard process because, by nature, is human intensive and domain specific. In many SW organizations, RE is performed in an ad hoc way by following specific practices which are not often defined explicitly. Even in case of more “mature” SW companies, which follow a pre-defined set of practices, a formal approach for measurement and assessment of these practices is usually not adopted, especially when these practices are applied to new development projects for a company. Second, SW development organizations are interested more in measuring the quality of the product produced than evaluating the quality of the process followed to produce the software product. Third, RE presents inherent dependencies with the consequent development phases and, thus, it is difficult to determine if a project failure is attributed to the poor quality of the RE process or to other defects derived from other stages of the development.

To meet these challenges in a practical way, in this WP our initial efforts will be devoted to identify a list of high-level requirements which constitute “good” RE quality attributes and success factors, across different software development projects. Then, we will move on to specify what data are required to quantify and measure these high-level success factors as well as what data collection procedures are necessary to be performed. Finally, WP1 will result in a detailed definition of low-level requirements (i.e., RE quality characteristics) and corresponding measures that will include: i) requirements contributing to the systematic definition, selection and prioritisation of the RE process, ii) requirements which influence positively the RE capability to manage changes and iii) requirements which have a positive effect on the capability of RE to define, estimate and manage risks and development costs. Quality Assessment Techniques, such as the Quality Function Deployment-QFD [37, 38] and the Goal-Question-Metrics Technique-GQM [39] will help to systematically apply to identify and prioritize these RE quality requirements and success factors.

In parallel with requirements analysis, our research will address approaches for improving the software process in the context of SW SMEs. We will primarily consider the needs of representative project examples from the SW companies involved in the project, with the main aim to support these companies to identify unsuccessful estimation/assessment practices and define actions to improve them. To achieve this goal we plan to exploit advantages from powerful estimation techniques such as the Bayesian Belief Networks (BBNs) [15, 40, 41]. BBNs can be helpful since they can provide: i) a way to represent project/process attributes and identify their interrelationships, ii) capabilities for multiple attribute estimations, iii) results indicating confidence of the estimation, iv) solutions that can be easily interpreted and confirmed by intuition, and v) a formal method that can be used alone or combined with expert judgment.

Members from the SPRINT SMEs research team have recently developed a methodology that supports process improvement by applying three phases, namely [41]: i) metrics collection of the current process, ii) selection, application and evaluation of formal process models expressed in BBN terms, and ii) specification of a new improved process. In SPRINT SMEs we will extend this methodology by considering an enhanced number of projects and consequently we will try to improve the estimation outputs. Therefore, the project will offer us the opportunity to gather data of on-going projects from the case companies in order to examine further the overall methodology and be able to provide more elaborate estimates.

Deliverables:

D2.1: Requirements Analysis for Software Process Assessment/Improvement Frameworks (M1-M8)

D2.2: Analysis Techniques for Supporting Software Process Assessment/Improvement (M1-M8)

WP2 Leader: Ioannis Stamelos

WP2 Members: V.C. Gerogiannis, G. Kakarontzas, D. Tselios, S. Bibi

Top